1 /*! \page faq Frequently Asked Questions
3 \htmlinclude .FAQ.doc.toc
5 \section faq_simgrid I'm new to SimGrid. I have some questions. Where should I start?
7 You are at the right place... Having a look to these
8 <a href="http://www.loria.fr/~quinson/blog/2010/06/28/Tutorial_at_HPCS/">the slides of the HPCS'10 tutorial</a>
9 (or to these <a href="http://graal.ens-lyon.fr/~alegrand/articles/slides_g5k_simul.pdf">ancient
10 slides</a>, or to these
11 <a href="http://graal.ens-lyon.fr/~alegrand/articles/Simgrid-Introduction.pdf">"obsolete" slides</a>)
12 may give you some insights on what SimGrid can help you to do and what
13 are its limitations. Then you definitely should read the \ref
14 MSG_examples. The \ref GRAS_tut can also help you.
16 If you are stuck at any point and if this FAQ cannot help you, please drop us a
17 mail to the user mailing list: <simgrid-user@lists.gforge.inria.fr>.
19 \subsection faq_interfaces What is the difference between MSG, SimDag, and GRAS? Do they serve the same purpose?
21 It depend on how you define "purpose", I guess ;)
23 They all allow you to build a prototype of application which you can run
24 within the simulator afterward. They all share the same simulation kernel,
25 which is the core of the SimGrid project. They differ by the way you express
28 With SimDag, you express your code as a collection of interdependent
29 parallel tasks. So, in this model, applications can be seen as a DAG of
30 tasks. This is the interface of choice for people wanting to port old
31 code designed for SimGrid v1 or v2 to the framework current version.
33 With both GRAS and MSG, your application is seen as a set of communicating
34 processes, exchanging data by the way of messages and performing computation
37 The difference between both is that MSG is somehow easier to use, but GRAS
38 is not limited to the simulator. Once you're done writing your GRAS code,
39 you can run your code both in the simulator or on a real platform. For this,
40 there is two implementations of the GRAS interface, one for simulation, one
41 for real execution. So, you just have to relink your code to chose one of
44 \subsection faq_generic First steps with SimGrid
46 If you decide to go for the MSG interface, please read carefully the
47 \ref MSG_examples. You'll find in \ref MSG_ex_master_slave a very
48 simple consisting of a master (that owns a bunch of tasks and
49 distributes them) , some slaves (that process tasks whenever they
50 receive one) and some forwarder agents (that simply pass the tasks
51 they receive to some slaves).
53 If you decide to go for the GRAS interface, you should definitively
54 read the \ref GRAS_tut. The first section constitutes an introduction
55 to the tool and presents the model we use. The second section
56 constitutes a complete step-by-step tutorial building a distributed
57 application from the beginning and exemplifying most of the GRAS
58 features in the process. The last section groups some HOWTOS
59 highlighting a given feature of the framework in a more concise way.
61 If you decide to go for another interface, I'm afraid your only sources
62 of information will be the source code and the mailing lists...
64 \subsection faq_visualization Visualizing and analyzing the results
66 It is sometime convenient to "see" how the agents are behaving. If you
67 like colors, you can use <tt>tools/MSG_visualization/colorize.pl </tt>
68 as a filter to your MSG outputs. It works directly with INFO. Beware,
69 INFO() prints on stderr. Do not forget to redirect if you want to
70 filter (e.g. with bash):
72 ./msg_test small_platform.xml small_deployment.xml 2>&1 | ../../tools/MSG_visualization/colorize.pl
75 We also have a more graphical output. Have a look at section \ref faq_tracing.
77 \subsection faq_C Argh! Do I really have to code in C?
79 Up until now, there is no binding for other languages. If you use C++,
80 you should be able to use the SimGrid library as a standard C library
81 and everything should work fine (simply <i>link</i> against this
82 library; recompiling SimGrid with a C++ compiler won't work and it
83 wouldn't help if you could).
85 In fact, we are currently working on Java bindings of MSG to allow
86 all the undergrad students of the world to use this tool. This is a
87 little more tricky than I would have expected, but the work is moving
88 fast forward [2006/05/13]. More languages are evaluated, but for now,
89 we do not feel a real demand for any other language. Please speak up!
91 \section faq_cmake Installing the SimGrid library with Cmake (since V3.4)
93 \subsection faq_intro Some generalitty
95 \subsubsection faq_intro1 What is Cmake?
97 CMake is a family of tools designed to build, test and package software. CMake is used to control the software compilation process using simple platform and compiler independent configuration files. CMake generates native makefiles and workspaces that can be used in the compiler environment of your choice. For more information see official web site <a href="http://www.cmake.org/">here</a>.
99 \subsubsection faq_intro2 Why cmake?
101 CMake permits to developers to compil projects on different plateforms. Then many tools are embedded like ctest for making test, a link to cdash for vizualise results but also test coverage and bug reports.
103 \subsubsection faq_intro3 What cmake need?
105 CMake needs some prerequists like :
110 \li c, c++ and java compiler regards to developers
111 \li ccmake for graphical used of CMake
112 \li cmake <a href="http://www.cmake.org/cmake/resources/software.html">(download page)</a>
115 \li cmake 2.8 <a href="http://www.cmake.org/cmake/resources/software.html">(download page)</a>
116 \li perl strawberry <a href="http://www.strawberryperl.com/download/5.12.2.0/strawberry-perl-5.12.2.0.msi">(download page)</a>
117 \li pcre-7.0 <a href="http://sourceforge.net/projects/gnuwin32/files/pcre/7.0/pcre-7.0.exe/download">(download page)</a>
118 \li git <a href="http://msysgit.googlecode.com/files/Git-1.7.4-preview20110204.exe">(download page)</a>
120 \subsubsection faq_cmakeoption1 Liste of options
123 "cmake -D[name]=[value] ... ./"
125 [name] enable_gtnets [value] ON/OFF or TRUE/FALSE or 1/0
126 enable_lua ON/OFF or TRUE/FALSE or 1/0
127 enable_compile_optimizations ON/OFF or TRUE/FALSE or 1/0
128 enable_compile_warnings ON/OFF or TRUE/FALSE or 1/0
129 enable_smpi ON/OFF or TRUE/FALSE or 1/0
130 enable_maintainer_mode ON/OFF or TRUE/FALSE or 1/0
131 enable_supernovae ON/OFF or TRUE/FALSE or 1/0
132 enable_tracing ON/OFF or TRUE/FALSE or 1/0
133 enable_coverage ON/OFF or TRUE/FALSE or 1/0
134 enable_memcheck ON/OFF or TRUE/FALSE or 1/0
135 enable_model-checking ON/OFF or TRUE/FALSE or 1/0
136 enable_debug ON/OFF or TRUE/FALSE or 1/0
137 enable_jedule ON/OFF or TRUE/FALSE or 1/0
138 enable_latency_bound_tracking ON/OFF or TRUE/FALSE or 1/0
139 enable_lib_static ON/OFF or TRUE/FALSE or 1/0
140 enable_pcre ON/OFF or TRUE/FALSE or 1/0
142 gtnets_path <path_to_gtnets_directory>
143 CMAKE_INSTALL_PREFIX <path_to_install_directory>
144 pipol_user <pipol_username>
147 \subsubsection faq_cmakeoption2 Options explaination
149 \li enable_gtnets : set to true implie that user wants to use gtnets.
151 \li enable_lua : set to true implie that user wants to add lua langage into simgrid compilation.
153 \li enable_compile_optimizations : add flags "-O3 -finline-functions -funroll-loops -fno-strict-aliasing"
155 \li enable_compile_warnings : add flags "-Wall -Wunused -Wmissing-prototypes -Wmissing-declarations -Wpointer-arith -Wchar-subscripts -Wcomment -Wformat -Wwrite-strings -Wno-unused-function -Wno-unused-parameter -Wno-strict-aliasing -Wno-format-nonliteral -Werror"
157 \li enable_smpi : Set to true if you want to use smpi lib. Actually on simgrid v3.4.1 Mac doesn't support lib smpi.
159 \li enable_maintainer_mode : set to true it remakes some files.
161 \li enable_supernovae : set to true make one file for each lib and compile with those generated files.
163 \li enable_tracing : To enable the generation of simulation traces for visualization.
165 \li enable_coverage : When set to true this option enable code coverage by setting -fprofile-arcs -ftest-coverage flags.
167 \li enable_memcheck : When set to true this option enable tests for memcheck.
169 \li enable_model-checking : Enable the model checking when set to true.
171 \li enable_debug : If enable_debug is set to 'off' Simgrid compil flag has '-DNDEBUG' option.
173 \li enable_jedule : To enable jedule mode, which creates visualizations of task schedules with Simdag.
175 \li enable_latency_bound_tracking : Set to on if you want to be warned when communications are limited by round trip time.
177 \li enable_lib_static : Enable generated Simgrid and smpi static libraries.
179 \li enable_pcre : Use or not the pcre lib for memory optimization.
181 \li custom_flags : If user wants to use a specific flag during compilation, give here.
183 \li gtnets_path : Path to gtnets install directory (ex /usr)
185 \li CMAKE_INSTALL_PREFIX : Path where are installed lib/ doc/ and include/ directories (ex /usr/local)
187 \li pipol_user : specify your pipol username if you want to use the pipol-remote command.
189 \subsubsection faq_cmakeoption3 Initialisation
191 Those options are initialized the first time you launch "cmake ." whithout specified option.
199 enable_compile_optimizations on
202 enable_compile_warnings off
203 enable_maintainer_mode off
206 enable_model-checking off
208 enable_latency_bound_tracking off
209 enable_lib_static off
210 CMAKE_INSTALL_PREFIX /usr/local
216 \subsubsection faq_cmakeoption4 Option's cache and how to reset?
218 When options have been set they are keep into a cache file named "CMakeCache.txt". So if you want
219 reset values you just delete this file located to the project directory.
221 \subsection faq_cmakecompilation Cmake compilation
223 \subsubsection faq_cmakecompilation1 With command line.
226 cmake -D[name]=[value] ... ./
232 Open the "Git Bash" window.
234 cmake -G"Unix Makefiles" -D[name]=[value] ... ./
238 \subsubsection faq_cmakecompilation2 With ccmake tool.
243 Then follow instructions.
245 \subsubsection faq_cmakecompilation2bis Build out of source.
247 As cmake generate many files used for compilation, we recommand to make a build directory.
248 For examples you can make :
251 "navarrop@caraja:~/Developments$ cd simgrid/"
252 "navarrop@caraja:~/Developments/simgrid$ mkdir build_directory"
253 "navarrop@caraja:~/Developments/simgrid$ cd build_directory/"
254 "navarrop@caraja:~/Developments/simgrid/build_directory$ cmake ../"
255 "navarrop@caraja:~/Developments/simgrid/build_directory$ make"
258 Or complety out of sources :
261 "navarrop@caraja:~/Developments$ mkdir build_dir"
262 "navarrop@caraja:~/Developments$ cd build_dir/"
263 "navarrop@caraja:~/Developments/build_dir$ cmake ../simgrid/"
264 "navarrop@caraja:~/Developments/build_dir$ make"
267 Those two kind of compilation permit to delete files created by compilation easier.
269 \subsubsection faq_cmakecompilation3 Resume of command line
273 cmake <path> configure the project
274 make build all targets
275 make VERBOSE=1 build all targets and print build command lines
276 make check test all targets and summarize
277 make dist make the distrib
278 make distcheck check the dist (make + make dist + make check)
279 make install install the project (doc/ bin/ lib/ include/)
280 make uninstall uninstall the project (doc/ bin/ lib/ include/)
281 make clean clean all targets
282 make simgrid_documentation Create simgrid documentation
285 When the project have been succesfully compiling and build you can make tests.
289 ctest launch only tests
291 ctest -D Continuous(Start|Update|Configure|Build)
292 ctest -D Continuous(Test|Coverage|MemCheck|Submit)
293 ctest -D Experimental
294 ctest -D Experimental(Start|Update|Configure|Build)
295 ctest -D Experimental(Test|Coverage|MemCheck|Submit)
297 ctest -D Nightly(Start|Update|Configure|Build)
298 ctest -D Nightly(Test|Coverage|MemCheck|Submit)
299 ctest -D NightlyMemoryCheck
302 If you want to test before make a commit you can simply make "ctest -D Experimental" and then you can visualize results submitted into Cdash. <a href="http://cdash.inria.fr/CDash/index.php?project=Simgrid">(Go to Cdash site)</a>.
304 \subsection faq_cmakeinstall How to install with cmake?
306 \subsubsection faq_cmakeinstall1 From svn.
310 cmake -Denable_maintainer_mode=on -DCMAKE_INSTALL_PREFIX=/home/navarrop/Bureau/install_simgrid ./
318 cmake -G"Unix Makefiles" -DCMAKE_INSTALL_PREFIX=C:\simgrid_install ./
323 \subsubsection faq_cmakeinstall2 From a distrib
326 For version 3.4.1 and 3.4
327 cmake -Dprefix=/home/navarrop/Bureau/install_simgrid ./
331 cmake -DCMAKE_INSTALL_PREFIX=/home/navarrop/Bureau/install_simgrid ./
336 \subsection faq_cmakeWHATisInstall What is installed by cmake?
338 \subsubsection faq_cmakeWHATisInstallBIN CMAKE_INSTALL_PREFIX/bin
350 \subsubsection faq_cmakeWHATisInstallDOC CMAKE_INSTALL_PREFIX/doc
355 \subsubsection faq_cmakeWHATisInstallINCLUDE CMAKE_INSTALL_PREFIX/include
371 \subsubsection faq_cmakeWHATisInstallLIB CMAKE_INSTALL_PREFIX/lib
376 libsimgrid.so -> libsimgrid.so.3.5
377 libgras.so -> libgras.so.3.5
378 libsmpi.so -> libsmpi.so.3.5
379 lua/5.1/simgrid.so -> ../../libsimgrid.so
380 ruby/1.9.0/x86_64-linux/libsimgrid.so -> ../../../libsimgrid.so
381 ruby/1.9.0/x86_64-linux/simgrid.rb
383 \subsection faq_cmakehowto How to modified sources files for developers
385 \subsubsection faq_cmakehowto1 Add an executable or examples.
387 If you want make an executable you have to create a CMakeList.txt to the src directory.
388 You must specified where to create the executable, source list, dependencies and the name of the binary.
391 cmake_minimum_required(VERSION 2.6)
393 set(EXECUTABLE_OUTPUT_PATH "./")
394 set(LIBRARY_OUTPUT_PATH "${CMAKE_HOME_DIRECTORY}/lib")
396 add_executable(get_sender get_sender.c) #add_executable(<name_of_target> <src list>)
398 ### Add definitions for compile
399 target_link_libraries(get_sender simgrid m pthread) #target_link_libraries(<name_of_targe> <dependencies>)
402 Then you have to modified <project/directory>/buildtools/Cmake/MakeExeLib.cmake and add
405 add_subdirectory(${CMAKE_HOME_DIRECTORY}/<path_where_is_CMakeList.txt>)
408 \subsubsection faq_cmakehowto2 Delete/add sources to lib.
410 If you want modified, add or delete source files from a library you have to edit <project/directory>/buildtools/Cmake/DefinePackages.cmake
414 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/MsgException.java
415 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/JniException.java
416 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/NativeException.java
417 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/HostNotFoundException.java
418 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/ProcessNotFoundException.java
419 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Msg.java
420 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Process.java
421 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Host.java
422 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Task.java
423 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/MsgNative.java
424 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/ApplicationHandler.java
425 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Sem.java
429 \subsubsection faq_cmakehowto3 Add test
431 If you want modified, add or delete tests you have to edit <project/directory>/buildtools/Cmake/AddTests.cmake
432 with this function : ADD_TEST(<name> <bin> <ARGS>)
435 add_test(test-simdag-1 ${CMAKE_HOME_DIRECTORY}/testsuite/simdag/sd_test --cfg=path:${CMAKE_HOME_DIRECTORY}/testsuite/simdag small_platform_variable.xml)
438 \subsection faq_PIPOL Pipol-remote
440 Now we offer the possibility to test your local sources on pipol platforms before a commit. Of course you have to be user of pipol <a href="https://pipol.inria.fr/users/">(Account request)</a> cause you need to give your pipol_username to cmake. Here is a list of available systems :
442 amd64_kvm-linux-debian-lenny
443 amd64_kvm-linux-debian-testing
445 amd64-linux-centos-5.dd.gz
446 amd64-linux-debian-etch.dd.gz
447 amd64-linux-debian-lenny.dd.gz
448 amd64-linux-debian-testing.dd.gz
449 amd64-linux-fedora-core10.dd.gz
450 amd64-linux-fedora-core11.dd.gz
451 amd64-linux-fedora-core12.dd.gz
452 amd64-linux-fedora-core13.dd.gz
453 amd64-linux-fedora-core7.dd.gz
454 amd64-linux-fedora-core8.dd.gz
455 amd64-linux-fedora-core9.dd.gz
456 amd64-linux-mandriva-2007_springs_powerpack.dd.gz
457 amd64-linux-mandriva-2009_powerpack.dd.gz
458 amd64-linux-opensuse-11.dd.gz
459 amd64-linux-redhatEL-5.0.dd.gz
460 amd64-linux-suse-LES10.dd.gz
461 amd64-linux-ubuntu-feisty.dd.gz
462 amd64-linux-ubuntu-hardy.dd.gz
463 amd64-linux-ubuntu-intrepid.dd.gz
464 amd64-linux-ubuntu-jaunty.dd.gz
465 amd64-linux-ubuntu-karmic.dd.gz
466 amd64-linux-ubuntu-lucid.dd.gz
467 amd64-unix-freebsd-7.dd.gz
468 amd64-windows-server-2003-64bits.dd.gz
469 amd64-windows-server-2008-64bits.dd.gz
470 i386_kvm-linux-debian-lenny
471 i386_kvm-linux-debian-testing
472 i386_kvm-linux-fedora-core13
473 i386_kvm-windows-xp-pro-sp3
474 i386-linux-centos-5.dd.gz
475 i386-linux-debian-etch.dd.gz
476 i386-linux-debian-lenny.dd.gz
477 i386-linux-debian-testing.dd.gz
478 i386-linux-fedora-core10.dd.gz
479 i386-linux-fedora-core11.dd.gz
480 i386-linux-fedora-core12.dd.gz
481 i386-linux-fedora-core13.dd.gz
482 i386-linux-fedora-core7.dd.gz
483 i386-linux-fedora-core8.dd.gz
484 i386-linux-fedora-core9.dd.gz
485 i386-linux-mandriva-2007_springs_powerpack.dd.gz
486 i386-linux-mandriva-2009_powerpack.dd.gz
487 i386-linux-opensuse-11.dd.gz
488 i386-linux-redhatEL-5.0.dd.gz
489 i386-linux-suse-LES10.dd.gz
490 i386-linux-ubuntu-feisty.dd.gz
491 i386-linux-ubuntu-hardy.dd.gz
492 i386-linux-ubuntu-intrepid.dd.gz
493 i386-linux-ubuntu-jaunty.dd.gz
494 i386-linux-ubuntu-karmic.dd.gz
495 i386-linux-ubuntu-lucid.dd.gz
496 i386_mac-mac-osx-server-leopard.dd.gz
497 i386-unix-freebsd-7.dd.gz
498 i386-unix-opensolaris-10.dd.gz
499 i386-unix-opensolaris-11.dd.gz
500 i386-unix-solaris-10.dd.gz
501 ia64-linux-debian-lenny.dd
502 ia64-linux-fedora-core9.dd
503 ia64-linux-redhatEL-5.0.dd
504 x86_64_mac-mac-osx-server-snow-leopard.dd.gz
505 x86_mac-mac-osx-server-snow-leopard.dd.gz
508 Two kind of uses are possible :
510 This command copy your source and execute a configure then a build and finish with tests.
511 bob@caraja:~/Developments/simgrid/tmp_build$ make <name_of_image>
513 This command copy your source and execute a \"ctest -D Experimental\" and submit the result to cdash.
514 bob@caraja:~/Developments/simgrid/tmp_build$ make <name_of_image>_experimental
516 All commands are resumed with :
518 bob@caraja:~/Developments/simgrid/tmp_build$ make pipol_experimental_list_images
519 bob@caraja:~/Developments/simgrid/tmp_build$ make pipol_test_list_images
522 \section faq_installation Installing the SimGrid library with Autotools (valid until V3.3.4)
524 Many people have been asking me questions on how to use SimGrid. Quite
525 often, the questions were not really about SimGrid but on the
526 installation process. This section is intended to help people that are
527 not familiar with compiling C files under UNIX. If you follow these
528 instructions and still have some troubles, drop an e-mail to
529 <simgrid-user@lists.gforge.inria.fr>.
531 \subsection faq_compiling Compiling SimGrid from a stable archive
533 First of all, you need to download the latest version of SimGrid from
534 <a href="http://gforge.inria.fr/frs/?group_id=12">here</a>.
535 Suppose you have uncompressed SimGrid in some temporary location of
536 your home directory (say <tt>/home/joe/tmp/simgrid-3.0.1 </tt>). The
537 simplest way to use SimGrid is to install it in your home
538 directory. Change your directory to
539 <tt>/home/joe/tmp/simgrid-3.0.1</tt> and type
542 ./configure --prefix=$HOME
547 If at some point, something fails, check the section \ref faq_trouble_compil .
548 If it does not help, you can report this problem to the
549 list but, please, avoid sending a laconic mail like "There is a problem. Is it
550 okay?". Send the config.log file which is automatically generated by
551 configure. Try to capture both the standard output and the error output of the
552 <tt>make</tt> command with <tt>script</tt>. There is no way for us to help you
553 without the relevant bits of information.
555 Now, the following directory should have been created :
557 \li <tt>/home/joe/doc/simgrid/html/</tt>
558 \li <tt>/home/joe/lib/</tt>
559 \li <tt>/home/joe/include/</tt>
561 SimGrid is not a binary, it is a library. Both a static and a dynamic
562 version are available. Here is what you can find if you try a <tt>ls
565 \verbatim libsimgrid.a libsimgrid.la libsimgrid.so libsimgrid.so.0 libsimgrid.so.0.0.1
568 Thus, there is two ways to link your program with SimGrid:
569 \li Either you use the static version, e.g
570 \verbatim gcc libsimgrid.a -o MainProgram MainProgram.c
572 In this case, all the SimGrid functions are directly
573 included in <tt>MainProgram</tt> (hence a bigger binary).
574 \li Either you use the dynamic version (the preferred method)
575 \verbatim gcc -lsimgrid -o MainProgram MainProgram.c
577 In this case, the SimGrid functions are not included in
578 <tt>MainProgram</tt> and you need to set your environment
579 variable in such a way that <tt>libsimgrid.so</tt> will be
580 found at runtime. This can be done by adding the following
581 line in your .bashrc (if you use bash and if you have
582 installed the SimGrid libraries in your home directory):
583 \verbatim export LD_LIBRARY_PATH=$HOME/lib/:$LD_LIBRARY_PATH
586 \subsection faq_compiling_java Java bindings don't get compiled
588 The configure script detects automatically whether you have the
589 softwares needed to use the Java bindings or not. At the end of the
590 configure, you can see the configuration picked by the script, which
591 should look similar to
592 \verbatim Configuration of package simgrid' (version 3.3.4-svn) on
595 Compiler: gcc (version: )
597 CFlags: -O3 -finline-functions -funroll-loops -fno-strict-aliasing -Wall -Wunused -Wmissing-prototypes -Wmissing-declarations -Wpointer-arith -Wchar-subscripts -Wcomment -Wformat -Wwrite-strings -Wno-unused-function -Wno-unused-parameter -Wno-strict-aliasing -Wno-format-nonliteral -Werror -g3
601 Context backend: ucontext
608 In this example, Java backends won't be compiled.
610 On Debian-like systems (which includes ubuntu), you need the following
611 packages: sun-java6-jdk libgcj10-dev. If you cannot find the
612 libgcj10-dev, try another version, like libgcj9-dev (on Ubuntu before
613 9.10) or libgcj11-dev (not released yet, but certainly one day).
614 Please note that you need to activate the contrib and non-free
615 repositories in Debian, and the universe ones in Ubuntu. Java comes at
618 \subsection faq_compiling_snapshoot SimGrid development snapshots
620 We have very high standards on software quality, and we are reluctant releasing
621 a stable release as long as there is still some known bug in the code base. In
622 addition, we added quite an extensive test base, making sure that we correctly
623 test the most important parts of the tool.
625 As an unfortunate conclusion, there may be some time between the stable
626 releases. If you want to benefit from the most recent features we introduced,
627 but don't want to take the risk of an untested version from the SVN, then
628 development snapshots are done for you.
630 These are pre-releases of SimGrid that still fail some tests about features
631 that almost nobody use, or on platforms not being in our core target (which is
632 Linux, Mac, other Unixes and Windows, from the most important to the less
633 one). That means that using this development releases should be safe for most
636 These archives can be found on
637 <a href="http://www.loria.fr/~quinson/Research/SimGrid/">this web page</a>. Once you
638 got the lastest archive, you can compile it just like any archive (see above).
640 \subsection faq_compiling_svn Compiling SimGrid from the SVN
642 The project development takes place in the SVN, where all changes are
643 committed when they happen. Then every once in a while, we make sure that the
644 code quality meets our standard and release an archive from the code in the
645 SVN. We afterward go back to the development in the SVN. So, if you need a
646 recently added feature and can afford some little problem with the stability
647 of the lastest features, you may want to use the SVN version instead of a
650 For that, you first need to get the "simgrid" module from
651 <a href="http://gforge.inria.fr/scm/?group_id=12">here</a>.
653 You won't find any <tt>configure</tt> and a few other things
654 (<tt>Makefile.in</tt>'s, documentation, ...) will be missing as well. The
655 reason for that is that all these files have to be regenerated using the
656 latest versions of <tt>autoconf</tt>, <tt>libtool</tt>, <tt>automake</tt>
657 (>1.9) and <tt>doxygen</tt> (>1.4). To generate the <tt>configure</tt> and
658 the <tt>Makefile.in</tt>'s, you just have to launch the <tt>bootstrap</tt>
659 command that resides in the top of the source tree. Then just follow the
660 instructions of Section \ref faq_compiling.
662 We insist on the fact that you really need the latest versions of
663 autoconf, automake and libtool. Doing this step on exotic architectures/systems
664 (i.e. anything different from a recent linux distribution) may be
665 ... uncertain. If you need to compile the SVN version on a machine where all these
666 dependencies are not met, the easiest is to do <tt>make dist</tt> in the SVN
667 directory of another machine where all dependencies are met. It will create an
668 archive you may deploy on other sites just as a regular stable release.
670 In summary, the following commands will checkout the SVN, regenerate the
671 configure script and friends, configure SimGrid and build it.
673 \verbatim svn checkout svn://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk simgrid
676 ./configure --enable-maintainer-mode --prefix=<where to install SimGrid>
679 Then, if you want to install SimGrid on the current box, just do:
680 \verbatim make install \endverbatim
682 If you want to build an snapshot of the SVN to deploy it on another box (for
683 example because the other machine don't have the autotools), do:
684 \verbatim make dist \endverbatim
686 Moreover, you should never call the autotools manually since you must run
687 them in a specific order with specific arguments. Most of the times, the
688 makefiles will automatically call the tools for you. When it's not possible
689 (such as the first time you checkout the SVN), use the ./bootstrap command
690 to call them explicitly.
693 \subsection faq_setting_MSG Setting up your own MSG code
695 Do not build your simulator by modifying the SimGrid examples. Go
696 outside the SimGrid source tree and create your own working directory
697 (say <tt>/home/joe/SimGrid/MyFirstScheduler/</tt>).
699 Suppose your simulation has the following structure (remember it is
700 just an example to illustrate a possible way to compile everything;
701 feel free to organize it as you want).
703 \li <tt>sched.h</tt>: a description of the core of the
704 scheduler (i.e. which functions are can be used by the
705 agents). For example we could find the following functions
706 (master, forwarder, slave).
708 \li <tt>sched.c</tt>: a C file including <tt>sched.h</tt> and
709 implementing the core of the scheduler. Most of these
710 functions use the MSG functions defined in section \ref
713 \li <tt>masterslave.c</tt>: a C file with the main function, i.e.
714 the MSG initialization (MSG_global_init()), the platform
715 creation (e.g. with MSG_create_environment()), the
716 deployment phase (e.g. with MSG_function_register() and
717 MSG_launch_application()) and the call to
720 To compile such a program, we suggest to use the following
721 Makefile. It is a generic Makefile that we have used many times with
722 our students when we teach the C language.
726 masterslave: masterslave.o sched.o
728 INSTALL_PATH = $$HOME
730 PEDANTIC_PARANOID_FREAK = -O0 -Wshadow -Wcast-align \
731 -Waggregate-return -Wmissing-prototypes -Wmissing-declarations \
732 -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations \
733 -Wmissing-noreturn -Wredundant-decls -Wnested-externs \
734 -Wpointer-arith -Wwrite-strings -finline-functions
735 REASONABLY_CAREFUL_DUDE = -Wall
736 NO_PRAYER_FOR_THE_WICKED = -w -O2
737 WARNINGS = $(REASONABLY_CAREFUL_DUDE)
738 CFLAGS = -g $(WARNINGS)
740 INCLUDES = -I$(INSTALL_PATH)/include
741 DEFS = -L$(INSTALL_PATH)/lib/
742 LDADD = -lm -lsimgrid
746 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) $^ $(LIBS) $(LDADD) -o $@
749 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) -c -o $@ $<
752 rm -f $(BIN_FILES) *.o *~
758 The first two lines indicates what should be build when typing make
759 (<tt>masterslave</tt>) and of which files it is to be made of
760 (<tt>masterslave.o</tt> and <tt>sched.o</tt>). This makefile assumes
761 that you have set up correctly your <tt>LD_LIBRARY_PATH</tt> variable
762 (look, there is a <tt>LDADD = -lm -lsimgrid</tt>). If you prefer using
763 the static version, remove the <tt>-lsimgrid</tt> and add a
764 <tt>$(INSTALL_PATH)/lib/libsimgrid.a</tt> on the next line, right
765 after the <tt>LIBS = </tt>.
767 More generally, if you have never written a Makefile by yourself, type
768 in a terminal : <tt>info make</tt> and read the introduction. The
769 previous example should be enough for a first try but you may want to
770 perform some more complex compilations...
772 \subsection faq_setting_GRAS Setting up your own GRAS code
774 If you use the GRAS interface instead of the MSG one, then previous section
775 is not the better source of information. Instead, you should check the GRAS
776 tutorial in general, and the \ref GRAS_tut_tour_setup in particular.
778 \section faq_howto Feature related questions
780 \subsection faq_MIA "Could you please add (your favorite feature here) to SimGrid?"
782 Here is the deal. The whole SimGrid project (MSG, SURF, GRAS, ...) is
783 meant to be kept as simple and generic as possible. We cannot add
784 functions for everybody's needs when these functions can easily be
785 built from the ones already in the API. Most of the time, it is
786 possible and when it was not possible we always have upgraded the API
787 accordingly. When somebody asks us a question like "How to do that?
788 Is there a function in the API to simply do this?", we're always glad
789 to answer and help. However if we don't need this code for our own
790 need, there is no chance we're going to write it... it's your job! :)
791 The counterpart to our answers is that once you come up with a neat
792 implementation of this feature (task duplication, RPC, thread
793 synchronization, ...), you should send it to us and we will be glad to
794 add it to the distribution. Thus, other people will take advantage of
795 it (and we don't have to answer this question again and again ;).
797 You'll find in this section a few "Missing In Action" features. Many
798 people have asked about it and we have given hints on how to simply do
799 it with MSG. Feel free to contribute...
801 \subsection faq_MIA_MSG MSG features
803 \subsubsection faq_MIA_examples I want some more complex MSG examples!
805 Many people have come to ask me a more complex example and each time,
806 they have realized afterward that the basics were in the previous three
809 Of course they have often been needing more complex functions like
810 MSG_process_suspend(), MSG_process_resume() and
811 MSG_process_isSuspended() (to perform synchronization), or
812 MSG_task_Iprobe() and MSG_process_sleep() (to avoid blocking
813 receptions), or even MSG_process_create() (to design asynchronous
814 communications or computations). But the examples are sufficient to
817 We know. We should add some more examples, but not really some more
818 complex ones... We should add some examples that illustrate some other
819 functionalists (like how to simply encode asynchronous
820 communications, RPC, process migrations, thread synchronization, ...)
821 and we will do it when we will have a little bit more time. We have
822 tried to document the examples so that they are understandable. Tell
823 us if something is not clear and once again feel free to participate!
826 \subsubsection faq_MIA_taskdup Missing in action: MSG Task duplication/replication
828 There is no task duplication in MSG. When you create a task, you can
829 process it or send it somewhere else. As soon as a process has sent
830 this task, he doesn't have this task anymore. It's gone. The receiver
831 process has got the task. However, you could decide upon receiving to
832 create a "copy" of a task but you have to handle by yourself the
833 semantic associated to this "duplication".
835 As we already told, we prefer keeping the API as simple as
836 possible. This kind of feature is rather easy to implement by users
837 and the semantic you associate really depends on people. Having a
838 *generic* task duplication mechanism is not that trivial (in
839 particular because of the data field). That is why I would recommand
840 that you write it by yourself even if I can give you advice on how to
843 You have the following functions to get informations about a task:
844 MSG_task_get_name(), MSG_task_get_compute_duration(),
845 MSG_task_get_remaining_computation(), MSG_task_get_data_size(),
846 and MSG_task_get_data().
848 You could use a dictionary (#xbt_dict_t) of dynars (#xbt_dynar_t). If
849 you still don't see how to do it, please come back to us...
851 \subsubsection faq_MIA_asynchronous I want to do asynchronous communications in MSG
853 In the past (version <= 3.4), there was no function to perform asynchronous communications.
854 It could easily be implemented by creating new process when needed though. Since version 3.5,
855 we have introduced the following functions:
864 We refer you to the description of these functions for more details on their usage as well
865 as to the exemple section on \ref MSG_ex_asynchronous_communications.
867 \subsubsection faq_MIA_thread_synchronization I need to synchronize my MSG processes
869 You obviously cannot use pthread_mutexes of pthread_conds since we handle every
870 scheduling related decision within SimGrid.
872 In the past (version <=3.3.4) you could do it by playing with
873 MSG_process_suspend() and MSG_process_resume() or with fake communications (using MSG_task_get(),
874 MSG_task_put() and MSG_task_Iprobe()).
876 Since version 3.4, you can use classical synchronization structures. See page \ref XBT_synchro or simply check in
877 include/xbt/synchro_core.h.
879 \subsubsection faq_MIA_host_load Where is the get_host_load function hidden in MSG?
881 There is no such thing because its semantic wouldn't be really
882 clear. Of course, it is something about the amount of host throughput,
883 but there is as many definition of "host load" as people asking for
884 this function. First, you have to remember that resource availability
885 may vary over time, which make any load notion harder to define.
887 It may be instantaneous value or an average one. Moreover it may be only the
888 power of the computer, or may take the background load into account, or may
889 even take the currently running tasks into account. In some SURF models,
890 communications have an influence on computational power. Should it be taken
893 First of all, it's near to impossible to predict the load beforehands in the
894 simulator since it depends on too much parameters (background load
895 variation, bandwidth sharing algorithmic complexity) some of them even being
896 not known beforehands (other task starting at the same time). So, getting
897 this information is really hard (just like in real life). It's not just that
898 we want MSG to be as painful as real life. But as it is in some way
899 realistic, we face some of the same problems as we would face in real life.
901 How would you do it for real? The most common option is to use something
902 like NWS that performs active probes. The best solution is probably to do
903 the same within MSG, as in next code snippet. It is very close from what you
904 would have to do out of the simulator, and thus gives you information that
905 you could also get in real settings to not hinder the realism of your
909 double get_host_load() {
910 m_task_t task = MSG_task_create("test", 0.001, 0, NULL);
911 double date = MSG_get_clock();
913 MSG_task_execute(task);
914 date = MSG_get_clock() - date;
915 MSG_task_destroy(task);
920 Of course, it may not match your personal definition of "host load". In this
921 case, please detail what you mean on the mailing list, and we will extend
922 this FAQ section to fit your taste if possible.
924 \subsubsection faq_MIA_communication_time How can I get the *real* communication time?
926 Communications are synchronous and thus if you simply get the time
927 before and after a communication, you'll only get the transmission
928 time and the time spent to really communicate (it will also take into
929 account the time spent waiting for the other party to be
930 ready). However, getting the *real* communication time is not really
931 hard either. The following solution is a good starting point.
936 m_task_t task = MSG_task_create("Task", task_comp_size, task_comm_size,
937 calloc(1,sizeof(double)));
938 *((double*) task->data) = MSG_get_clock();
939 MSG_task_put(task, slaves[i % slaves_count], PORT_22);
940 XBT_INFO("Send completed");
945 m_task_t task = NULL;
948 time1 = MSG_get_clock();
949 a = MSG_task_get(&(task), PORT_22);
950 time2 = MSG_get_clock();
951 if(time1<*((double *)task->data))
952 time1 = *((double *) task->data);
953 XBT_INFO("Communication time : \"%f\" ", time2-time1);
955 MSG_task_destroy(task);
960 \subsection faq_MIA_SimDag SimDag related questions
962 \subsubsection faq_SG_comm Implementing communication delays between tasks.
964 A classic question of SimDag newcomers is about how to express a
965 communication delay between tasks. The thing is that in SimDag, both
966 computation and communication are seen as tasks. So, if you want to
967 model a data dependency between two DAG tasks t1 and t2, you have to
968 create 3 SD_tasks: t1, t2 and c and add dependencies in the following
972 SD_task_dependency_add(NULL, NULL, t1, c);
973 SD_task_dependency_add(NULL, NULL, c, t2);
976 This way task t2 cannot start before the termination of communication c
977 which in turn cannot start before t1 ends.
979 When creating task c, you have to associate an amount of data (in bytes)
980 corresponding to what has to be sent by t1 to t2.
982 Finally to schedule the communication task c, you have to build a list
983 comprising the workstations on which t1 and t2 are scheduled (w1 and w2
984 for example) and build a communication matrix that should look like
987 \subsubsection faq_SG_DAG How to implement a distributed dynamic scheduler of DAGs.
989 Distributed is somehow "contagious". If you start making distributed
990 decisions, there is no way to handle DAGs directly anymore (unless I
991 am missing something). You have to encode your DAGs in term of
992 communicating process to make the whole scheduling process
993 distributed. Here is an example of how you could do that. Assume T1
994 has to be done before T2.
997 int your_agent(int argc, char *argv[] {
999 T1 = MSG_task_create(...);
1000 T2 = MSG_task_create(...);
1004 if(cond) MSG_task_execute(T1);
1006 if((MSG_task_get_remaining_computation(T1)=0.0) && (you_re_in_a_good_mood))
1007 MSG_task_execute(T2)
1009 /* do something else */
1015 If you decide that the distributed part is not that much important and that
1016 DAG is really the level of abstraction you want to work with, then you should
1017 give a try to \ref SD_API.
1019 \subsection faq_MIA_generic Generic features
1021 \subsubsection faq_more_processes Increasing the amount of simulated processes
1023 Here are a few tricks you can apply if you want to increase the amount
1024 of processes in your simulations.
1026 - <b>A few thousands of simulated processes</b> (soft tricks)\n
1027 SimGrid can use either pthreads library or the UNIX98 contextes. On
1028 most systems, the number of pthreads is limited and then your
1029 simulation may be limited for a stupid reason. This is especially
1030 true with the current linux pthreads, and I cannot get more than
1031 2000 simulated processes with pthreads on my box. The UNIX98
1032 contexts allow me to raise the limit to 25,000 simulated processes
1034 The <tt>--with-context</tt> option of the <tt>./configure</tt>
1035 script allows you to choose between UNIX98 contextes
1036 (<tt>--with-context=ucontext</tt>) and the pthread version
1037 (<tt>--with-context=pthread</tt>). The default value is ucontext
1038 when the script detect a working UNIX98 context implementation. On
1039 Windows boxes, the provided value is discarded and an adapted
1040 version is picked up.\n\n
1041 We experienced some issues with contextes on some rare systems
1042 (solaris 8 and lower or old alpha linuxes comes to mind). The main
1043 problem is that the configure script detect the contextes as being
1044 functional when it's not true. If you happen to use such a system,
1045 switch manually to the pthread version, and provide us with a good
1046 patch for the configure script so that it is done automatically ;)
1048 - <b>Hundred thousands of simulated processes</b> (hard-core tricks)\n
1049 As explained above, SimGrid can use UNIX98 contextes to represent
1050 and handle the simulated processes. Thanks to this, the main
1051 limitation to the number of simulated processes becomes the
1052 available memory.\n\n
1053 Here are some tricks I had to use in order to run a token ring
1054 between 25,000 processes on my laptop (1Gb memory, 1.5Gb swap).\n
1055 - First of all, make sure your code runs for a few hundreds
1056 processes before trying to push the limit. Make sure it's
1057 valgrind-clean, ie that valgrind does not report neither memory
1058 error nor memory leaks. Indeed, numerous simulated processes
1059 result in *fat* simulation hindering debugging.
1060 - It was really boring to write 25,000 entries in the deployment
1061 file, so I wrote a little script
1062 <tt>examples/gras/mutual_exclusion/simple_token/make_deployment.pl</tt>, which you may
1063 want to adapt to your case. You could also think about hijacking
1064 the SURFXML parser (have look at \ref faq_flexml_bypassing).
1065 - The deployment file became quite big, so I had to do what is in
1066 the FAQ entry \ref faq_flexml_limit
1067 - Each UNIX98 context has its own stack entry. As debugging this is
1068 quite hairly, the default value is a bit overestimated so that
1069 user don't get into trouble about this. You want to tune this
1070 size to increse the number of processes. This is the
1071 <tt>STACK_SIZE</tt> define in
1072 <tt>src/xbt/xbt_context_sysv.c</tt>, which is 128kb by default.
1073 Reduce this as much as you can, but be warned that if this value
1074 is too low, you'll get a segfault. The token ring example, which
1075 is quite simple, runs with 40kb stacks.
1076 - You may tweak the logs to reduce the stack size further. When
1077 logging something, we try to build the string to display in a
1078 char array on the stack. The size of this array is constant (and
1079 equal to XBT_LOG_BUFF_SIZE, defined in include/xbt/log/h). If the
1080 string is too large to fit this buffer, we move to a dynamically
1081 sized buffer. In which case, we have to traverse one time the log
1082 event arguments to compute the size we need for the buffer,
1083 malloc it, and traverse the argument list again to do the actual
1085 The idea here is to move XBT_LOG_BUFF_SIZE to 1, forcing the logs
1086 to use a dynamic array each time. This allows us to lower further
1087 the stack size at the price of some performance loss...\n
1088 This allowed me to run the reduce the stack size to ... 4k. Ie,
1089 on my 1Gb laptop, I can run more than 250,000 processes!
1091 \subsubsection faq_MIA_batch_scheduler Is there a native support for batch schedulers in SimGrid?
1093 No, there is no native support for batch schedulers and none is
1094 planned because this is a very specific need (and doing it in a
1095 generic way is thus very hard). However some people have implemented
1096 their own batch schedulers. Vincent Garonne wrote one during his PhD
1097 and put his code in the contrib directory of our SVN so that other can
1098 keep working on it. You may find inspiring ideas in it.
1100 \subsubsection faq_MIA_checkpointing I need a checkpointing thing
1102 Actually, it depends on whether you want to checkpoint the simulation, or to
1103 simulate checkpoints.
1105 The first one could help if your simulation is a long standing process you
1106 want to keep running even on hardware issues. It could also help to
1107 <i>rewind</i> the simulation by jumping sometimes on an old checkpoint to
1108 cancel recent calculations.\n
1109 Unfortunately, such thing will probably never exist in SG. One would have to
1110 duplicate all data structures because doing a rewind at the simulator level
1111 is very very hard (not talking about the malloc free operations that might
1112 have been done in between). Instead, you may be interested in the Libckpt
1113 library (http://www.cs.utk.edu/~plank/plank/www/libckpt.html). This is the
1114 checkpointing solution used in the condor project, for example. It makes it
1115 easy to create checkpoints (at the OS level, creating something like core
1116 files), and rerunning them on need.
1118 If you want to simulate checkpoints instead, it means that you want the
1119 state of an executing task (in particular, the progress made towards
1120 completion) to be saved somewhere. So if a host (and the task executing on
1121 it) fails (cf. #MSG_HOST_FAILURE), then the task can be restarted
1122 from the last checkpoint.\n
1124 Actually, such a thing does not exists in SimGrid either, but it's just
1125 because we don't think it is fundamental and it may be done in the user code
1126 at relatively low cost. You could for example use a watcher that
1127 periodically get the remaining amount of things to do (using
1128 MSG_task_get_remaining_computation()), or fragment the task in smaller
1131 \subsection faq_platform Platform building and Dynamic resources
1133 \subsubsection faq_platform_example Where can I find SimGrid platform files?
1135 There is several little examples in the archive, in the examples/msg
1136 directory. From time to time, we are asked for other files, but we
1137 don't have much at hand right now.
1139 You should refer to the Platform Description Archive
1140 (http://pda.gforge.inria.fr) project to see the other platform file we
1141 have available, as well as the Simulacrum simulator, meant to generate
1142 SimGrid platforms using all classical generation algorithms.
1144 \subsubsection faq_platform_alnem How can I automatically map an existing platform?
1146 We are working on a project called ALNeM (Application-Level Network
1147 Mapper) which goal is to automatically discover the topology of an
1148 existing network. Its output will be a platform description file
1149 following the SimGrid syntax, so everybody will get the ability to map
1150 their own lab network (and contribute them to the catalog project).
1151 This tool is not ready yet, but it move quite fast forward. Just stay
1154 \subsubsection faq_platform_synthetic Generating synthetic but realistic platforms
1156 The third possibility to get a platform file (after manual or
1157 automatic mapping of real platforms) is to generate synthetic
1158 platforms. Getting a realistic result is not a trivial task, and
1159 moreover, nobody is really able to define what "realistic" means when
1160 speaking of topology files. You can find some more thoughts on this
1162 <a href="http://graal.ens-lyon.fr/~alegrand/articles/Simgrid-Introduction.pdf">slides</a>.
1164 If you are looking for an actual tool, there we have a little tool to
1165 annotate Tiers-generated topologies. This perl-script is in
1166 <tt>tools/platform_generation/</tt> directory of the SVN. Dinda et Al.
1167 released a very comparable tool, and called it GridG.
1169 \subsubsection faq_SURF_multicore Modeling multi-core resources
1171 There is currently no native support for multi-core or SMP machines in
1172 SimGrid. We are currently working on it, but coming up with the right
1173 model is very hard: Cores share caches and bus to access memory and
1174 thus interfere with each others. Memory contention is a crucial
1175 component of multi-core modeling.
1177 In the meanwhile, some user-level tricks can reveal sufficient for
1178 you. For example, you may model each core by a CPU and add some very
1179 high speed links between them. This complicates a bit the user code
1180 since you have to remember that when you assign something to a (real)
1181 host, it can be any of the (fake) hosts representing the cores of a
1182 given machine. For that, you can use the prop tag of the XML files as
1183 follows. Your code should then look at the ‘machine’ property
1184 associated with each workstation, and run parallel tasks over all
1185 cores of the machine.
1188 <host id="machine0/core0" power="91500E6">
1189 <prop id="machine" value="machine0"/>
1190 <prop id="core" value="0"/>
1192 <host id="machine0/core1" power="91500E6">
1193 <prop id="machine" value="machine0"/>
1194 <prop id="core" value="1"/>
1200 \subsubsection faq_SURF_dynamic Modeling dynamic resource availability
1202 A nice feature of SimGrid is that it enables you to seamlessly have
1203 resources whose availability change over time. When you build a
1204 platform, you generally declare hosts like that:
1207 <host id="host A" power="100.00"/>
1210 If you want the availability of "host A" to change over time, the only
1211 thing you have to do is change this definition like that:
1214 <host id="host A" power="100.00" availability_file="trace_A.txt" state_file="trace_A_failure.txt"/>
1217 For hosts, availability files are expressed in fraction of available
1218 power. Let's have a look at what "trace_A.txt" may look like:
1227 At time 0, our host will deliver 100 flop/s. At time 11.0, it will
1228 deliver only 50 flop/s until time 20.0 where it will will start
1229 delivering 90 flop/s. Last at time 21.0 (20.0 plus the periodicity
1230 1.0), we'll be back to the beginning and it will deliver 100 flop/s.
1232 Now let's look at the state file:
1239 A negative value means "off" while a positive one means "on". At time
1240 1.0, the host is on. At time 1.0, it is turned off and at time 2.0, it
1241 is turned on again until time 12 (2.0 plus the periodicity 10.0). It
1242 will be turned on again at time 13.0 until time 23.0, and so on.
1244 Now, let's look how the same kind of thing can be done for network
1245 links. A usual declaration looks like:
1248 <link id="LinkA" bandwidth="10.0" latency="0.2"/>
1251 You have at your disposal the following options: bandwidth_file,
1252 latency_file and state_file. The only difference with hosts is that
1253 bandwidth_file and latency_file do not express fraction of available
1254 power but are expressed directly in bytes per seconds and seconds.
1256 \subsubsection faq_platform_multipath How to express multipath routing in platform files?
1258 It is unfortunately impossible to express the fact that there is more
1259 than one routing path between two given hosts. Let's consider the
1260 following platform file:
1263 <route src="A" dst="B">
1266 <route src="B" dst="C">
1269 <route src="A" dst="C">
1274 Although it is perfectly valid, it does not mean that data traveling
1275 from A to C can either go directly (using link 3) or through B (using
1276 links 1 and 2). It simply means that the routing on the graph is not
1277 trivial, and that data do not following the shortest path in number of
1278 hops on this graph. Another way to say it is that there is no implicit
1279 in these routing descriptions. The system will only use the routes you
1280 declare (such as <route src="A" dst="C"><link:ctn
1281 id="3"/></route>), without trying to build new routes by aggregating
1284 You are also free to declare platform where the routing is not
1285 symmetric. For example, add the following to the previous file:
1288 <route src="C" dst="A">
1294 This makes sure that data from C to A go through B where data from A
1295 to C go directly. Don't worry about realism of such settings since
1296 we've seen ways more weird situation in real settings (in fact, that's
1297 the realism of very regular platforms which is questionable, but
1298 that's another story).
1300 \subsubsection faq_flexml_bypassing Bypassing the XML parser with your own C functions
1302 So you want to bypass the XML files parser, uh? Maybe doing some parameter
1303 sweep experiments on your simulations or so? This is possible, and
1304 it's not even really difficult (well. Such a brutal idea could be
1305 harder to implement). Here is how it goes.
1307 For this, you have to first remember that the XML parsing in SimGrid is done
1308 using a tool called FleXML. Given a DTD, this gives a flex-based parser. If
1309 you want to bypass the parser, you need to provide some code mimicking what
1310 it does and replacing it in its interactions with the SURF code. So, let's
1311 have a look at these interactions.
1313 FleXML parser are close to classical SAX parsers. It means that a
1314 well-formed SimGrid platform XML file might result in the following
1317 - start "platform_description" with attribute version="2"
1318 - start "host" with attributes id="host1" power="1.0"
1320 - start "host" with attributes id="host2" power="2.0"
1322 - start "link" with ...
1324 - start "route" with ...
1325 - start "link:ctn" with ...
1328 - end "platform_description"
1330 The communication from the parser to the SURF code uses two means:
1331 Attributes get copied into some global variables, and a surf-provided
1332 function gets called by the parser for each event. For example, the event
1333 - start "host" with attributes id="host1" power="1.0"
1335 let the parser do something roughly equivalent to:
1337 strcpy(A_host_id,"host1");
1342 In SURF, we attach callbacks to the different events by initializing the
1343 pointer functions to some the right surf functions. Since there can be
1344 more than one callback attached to the same event (if more than one
1345 model is in use, for example), they are stored in a dynar. Example in
1346 workstation_ptask_L07.c:
1348 /* Adding callback functions */
1349 surf_parse_reset_parser();
1350 surfxml_add_callback(STag_surfxml_host_cb_list, &parse_cpu_init);
1351 surfxml_add_callback(STag_surfxml_prop_cb_list, &parse_properties);
1352 surfxml_add_callback(STag_surfxml_link_cb_list, &parse_link_init);
1353 surfxml_add_callback(STag_surfxml_route_cb_list, &parse_route_set_endpoints);
1354 surfxml_add_callback(ETag_surfxml_link_c_ctn_cb_list, &parse_route_elem);
1355 surfxml_add_callback(ETag_surfxml_route_cb_list, &parse_route_set_route);
1357 /* Parse the file */
1358 surf_parse_open(file);
1359 xbt_assert(!surf_parse(), "Parse error in %s", file);
1363 So, to bypass the FleXML parser, you need to write your own version of the
1364 surf_parse function, which should do the following:
1365 - Fill the A_<tag>_<attribute> variables with the wanted values
1366 - Call the corresponding STag_<tag>_fun function to simulate tag start
1367 - Call the corresponding ETag_<tag>_fun function to simulate tag end
1368 - (do the same for the next set of values, and loop)
1370 Then, tell SimGrid that you want to use your own "parser" instead of the stock one:
1372 surf_parse = surf_parse_bypass_environment;
1373 MSG_create_environment(NULL);
1374 surf_parse = surf_parse_bypass_application;
1375 MSG_launch_application(NULL);
1378 A set of macros are provided at the end of
1379 include/surf/surfxml_parse.h to ease the writing of the bypass
1380 functions. An example of this trick is distributed in the file
1381 examples/msg/masterslave/masterslave_bypass.c
1383 \subsection faq_simgrid_configuration Changing SimGrid's behavior
1385 A number of options can be given at runtime to change the default
1386 SimGrid behavior. In particular, you can change the default cpu and
1389 \subsubsection faq_simgrid_configuration_fullduplex Using Fullduplex
1391 Experimental fullduplex support is now available on the svn branch. In order to fullduple to work your platform must have two links for each pair
1392 of interconnected hosts, see an example here:
1394 simgrid_svn_sources/exemples/msg/gtnets/fullduplex-p.xml
1397 Using fullduplex support ongoing and incoming communication flows are
1398 treated independently for most models. The exception is the LV08 model which
1399 adds 0.05 of usage on the opposite direction for each new created flow. This
1400 can be useful to simulate some important TCP phenomena such as ack compression.
1402 Running a fullduplex example:
1404 cd simgrid_svn_sources/exemples/msg/gtnets
1405 ./gtnets fullduplex-p.xml fullduplex-d.xml --cfg=fullduplex:1
1412 \subsubsection faq_simgrid_configuration_gtnets Using GTNetS
1414 It is possible to use a packet-level network simulator
1415 instead of the default flow-based simulation. You may want to use such
1416 an approach if you have doubts about the validity of the default model
1417 or if you want to perform some validation experiments. At the moment,
1418 we support the GTNetS simulator (it is still rather experimental
1419 though, so leave us a message if you play with it).
1423 To enable GTNetS model inside SimGrid it is needed to patch the GTNetS simulator source code
1424 and build/install it from scratch
1427 - <b>Download and enter the recent downloaded GTNetS directory</b>
1430 svn checkout svn://scm.gforge.inria.fr/svn/simgrid/contrib/trunk/GTNetS/
1435 - <b>Use the following commands to unzip and patch GTNetS package to work within SimGrid.</b>
1438 unzip gtnets-current.zip
1439 tar zxvf gtnets-current-patch.tgz
1441 cat ../00*.patch | patch -p1
1444 - <b>OPTIONALLY</b> you can use a patch for itanium 64bit processor family.
1447 cat ../AMD64-FATAL-Removed-DUL_SIZE_DIFF-Added-fPIC-compillin.patch | patch -p1
1450 - <b>Compile GTNetS</b>
1452 Due to portability issues it is possible that GTNetS does not compile in your architecture. The patches furnished in SimGrid SVN repository are intended for use in Linux architecture only. Unfortunately, we do not have the time, the money, neither the manpower to guarantee GTNetS portability. We advice you to use one of GTNetS communication channel to get more help in compiling GTNetS.
1456 ln -sf Makefile.linux Makefile
1462 - <b>NOTE</b> A lot of warnings are expected but the application should compile
1463 just fine. If the makefile insists in compiling some QT libraries
1464 please try a make clean before asking for help.
1467 - <b>To compile optimized version</b>
1474 - <b>Installing GTNetS</b>
1476 It is important to put the full path of your libgtsim-xxxx.so file when creating the symbolic link. Replace < userhome > by some path you have write access to.
1479 ln -sf /<absolute_path>/gtnets_current/libgtsim-debug.so /<userhome>/usr/lib/libgtnets.so
1480 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/<userhome>/usr/lib/libgtnets.so
1481 mkdir /<userhome>/usr/include/gtnets
1482 cp -fr SRC/*.h /<userhome>/usr/include/gtnets
1486 - <b>Enable GTNetS support in SimGrid</b>
1488 In order to enable gtnets with simgrid you have to give where is gtnets. (path to \<gtnets_path\>/lib and \<gtnets_path\>/include)
1491 Since v3.4 (with cmake)
1492 cmake . -Dgtnets_path=/<userhome>/usr
1494 Until v3.4 (with autotools)
1495 ./configure --with-gtnets=/<userhome>/usr
1498 - <b>Once you have followed all the instructions for compiling and
1499 installing successfully you can activate this feature at
1500 runntime with the following options:</b>
1503 Since v3.4 (with cmake)
1508 Until v3.4 (with autotools)
1509 cd simgrid/example/msg/
1515 - <b>Or try the GTNetS model dogbone example with</b>
1518 gtnets/gtnets gtnets/onelink-p.xml gtnets/onelink-d.xml --cfg=network_model:GTNets
1522 A long version of this <a href="http://gforge.inria.fr/docman/view.php/12/6283/GTNetS HowTo.html">HowTo</a> it is available
1525 More about GTNetS simulator at <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/index.html">GTNetS Website</a>
1529 The patches provided by us worked successfully with GTNetS found
1530 <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/software/gtnets-current.zip">here</a>,
1531 dated from 12th June 2008. Due to the discontinuing development of
1532 GTNetS it is impossible to precise a version number. We STRONGLY recommend you
1533 to download and install the GTNetS version found in SimGrid repository as explained above.
1538 \subsubsection faq_simgrid_configuration_alternate_network Using alternative flow models
1540 The default simgrid network model uses a max-min based approach as
1541 explained in the research report
1542 <a href="ftp://ftp.ens-lyon.fr/pub/LIP/Rapports/RR/RR2002/RR2002-40.ps.gz">A Network Model for Simulation of Grid Application</a>.
1543 Other models have been proposed and implemented since then (see for example
1544 <a href="http://mescal.imag.fr/membres/arnaud.legrand/articles/simutools09.pdf">Accuracy Study and Improvement of Network Simulation in the SimGrid Framework</a>)
1545 and can be activated at runtime. For example:
1547 ./mycode platform.xml deployment.xml --cfg=workstation/model:compound --cfg=network/model:LV08 -cfg=cpu/model:Cas01
1550 Possible models for the network are currently "Constant", "CM02",
1551 "LegrandVelho", "GTNets", Reno", "Reno2", "Vegas". Others will
1552 probably be added in the future and many of the previous ones are
1553 experimental and are likely to disappear without notice... To know the
1554 list of the currently implemented models, you should use the
1555 --help-models command line option.
1558 ./masterslave_forwarder ../small_platform.xml deployment_masterslave.xml --help-models
1559 Long description of the workstation models accepted by this simulator:
1560 CLM03: Default workstation model, using LV08 and CM02 as network and CPU
1561 compound: Workstation model allowing you to use other network and CPU models
1562 ptask_L07: Workstation model with better parallel task modeling
1563 Long description of the CPU models accepted by this simulator:
1564 Cas01_fullupdate: CPU classical model time=size/power
1565 Cas01: Variation of Cas01_fullupdate with partial invalidation optimization of lmm system. Should produce the same values, only faster
1566 CpuTI: Variation of Cas01 with also trace integration. Should produce the same values, only faster if you use availability traces
1567 Long description of the network models accepted by this simulator:
1568 Constant: Simplistic network model where all communication take a constant time (one second)
1569 CM02: Realistic network model with lmm_solve and no correction factors
1570 LV08: Realistic network model with lmm_solve and these correction factors: latency*=10.4, bandwidth*=.92, S=8775
1571 Reno: Model using lagrange_solve instead of lmm_solve (experts only)
1572 Reno2: Model using lagrange_solve instead of lmm_solve (experts only)
1573 Vegas: Model using lagrange_solve instead of lmm_solve (experts only)
1576 \subsection faq_tracing Tracing Simulations for Visualization
1578 The trace visualization is widely used to observe and understand the behavior
1579 of parallel applications and distributed algorithms. Usually, this is done in a
1580 two-step fashion: the user instruments the application and the traces are
1581 analyzed after the end of the execution. The visualization itself can highlights
1582 unexpected behaviors, bottlenecks and sometimes can be used to correct
1583 distributed algorithms. The SimGrid team has instrumented the library
1584 in order to let users trace their simulations and analyze them. This part of the
1585 user manual explains how the tracing-related features can be enabled and used
1586 during the development of simulators using the SimGrid library.
1588 \subsubsection faq_tracing_howitworks How it works
1590 For now, the SimGrid library is instrumented so users can trace the <b>platform
1591 utilization</b> using the MSG, SimDAG and SMPI interface. This means that the tracing will
1592 register how much power is used for each host and how much bandwidth is used for
1593 each link of the platform. The idea with this type of tracing is to observe the
1594 overall view of resources utilization in the first place, especially the
1595 identification of bottlenecks, load-balancing among hosts, and so on.
1597 The idea of the tracing facilities is to give SimGrid users to possibility to
1598 classify MSG and SimDAG tasks by category, tracing the platform utilization
1599 (hosts and links) for each of the categories. For that,
1600 the tracing interface enables the declaration of categories and a function to
1601 mark a task with a previously declared category. <em>The tasks that are not
1602 classified according to a category are not traced</em>. Even if the user
1603 does not specify any category, the simulations can still be traced in terms
1604 of resource utilization by using a special parameter that is detailed below.
1606 \subsubsection faq_tracing_enabling Enabling using CMake
1608 With the sources of SimGrid, it is possible to enable the tracing
1609 using the parameter <b>-Denable_tracing=ON</b> when the cmake is executed.
1610 The section \ref faq_tracing_functions describes all the functions available
1611 when this Cmake options is activated. These functions will have no effect
1612 if SimGrid is configured without this option (they are wiped-out by the
1616 $ cmake -Denable_tracing=ON .
1620 \subsubsection faq_tracing_functions Tracing Functions
1622 \li <b>\c TRACE_category (const char *category)</b>: This function should be used
1623 to define a user category. The category can be used to differentiate the tasks
1624 that are created during the simulation (for example, tasks from server1,
1625 server2, or request tasks, computation tasks, communication tasks).
1626 All resource utilization (host power and link bandwidth) will be
1627 classified according to the task category. Tasks that do not belong to a
1628 category are not traced. The color for the category that is being declared
1629 is random (use next function to specify a color).
1631 \li <b>\c TRACE_category_with_color (const char *category, const char *color)</b>: Same
1632 as TRACE_category, but let user specify a color encoded as a RGB-like string with
1633 three floats from 0 to 1. So, to specify a red color, the user can pass "1 0 0" as
1634 color parameter. A light-gray color can be specified using "0.7 0.7 0.7" as color.
1636 \li <b>\c TRACE_msg_set_task_category (m_task_t task, const char *category)</b>:
1637 This function should be called after the creation of a MSG task, to define the
1638 category of that task. The first parameter \c task must contain a task that was
1639 created with the function \c MSG_task_create. The second parameter
1640 \c category must contain a category that was previously defined by the function
1643 \li <b>\c TRACE_sd_set_task_category (SD_task_t task, const char *category)</b>:
1644 This function should be called after the creation of a SimDAG task, to define the
1645 category of that task. The first parameter \c task must contain a task that was
1646 created with the function \c MSG_task_create. The second parameter
1647 \c category must contain a category that was previously defined by the function
1650 \li <b>\c TRACE_[host|link]_variable_declare (const char *variable)</b>:
1651 Declare a user variable that will be associated to host/link. A variable can
1652 be used to trace user variables such as the number of tasks in a server,
1653 the number of clients in an application (for hosts), and so on.
1655 \li <b>\c TRACE_[host|link]_variable_[set|add|sub] (const char *[host|link], const char *variable, double value)</b>:
1656 Set the value of a given user variable for a given host/link. The value
1657 of this variable is always associated to the host/link. The host/link
1658 parameters should be its name as the one listed in the platform file.
1660 \li <b>\c TRACE_[host|link]_variable_[set|add|sub]_with_time (double time, const char *[host|link], const char *variable, double value)</b>:
1661 Same as TRACE_[host|link]_variable_[set|add|sub], but let user specify
1662 the time used to trace it. Users can specify a time that is not the
1663 simulated clock time as defined by the core simulator. This allows
1664 a fine-grain control of time definition, but should be used with
1665 caution since the trace can be inconsistent if resource utilization
1666 traces are also traced.
1668 \li <b>\c TRACE_link_srcdst_variable_[set|add|sub] (const char *src, const char *dst, const char *variable, double value)</b>:
1669 Same as TRACE_link_variable_[set|add|sub], but now users specify a source and
1670 destination hosts (as the names from the platform file). The tracing library
1671 will get the corresponding route that connects those two hosts (src and dst) and
1672 [set|add|sub] the value's variable for all the links of the route.
1674 \li <b>\c TRACE_link_srcdst_variable_[set|add|sub]_with_time (double time, const char *src, const char *dst, const char *variable, double value)</b>:
1675 Same as TRACE_link_srcdst_variable_[set|add|sub], but user specify a time different from the simulated time.
1677 \subsubsection faq_tracing_options Tracing configuration Options
1679 These are the options accepted by the tracing system of SimGrid:
1684 Safe switch. It activates (or deactivates) the tracing system.
1685 No other tracing options take effect if this one is not activated.
1690 Register the simulation platform in the trace file.
1693 tracing/onelink_only
1695 By default, the tracing system uses all routes in the platform file
1696 to re-create a "graph" of the platform and register it in the trace file.
1697 This option let the user tell the tracing system to use only the routes
1698 that are composed with just one link.
1703 It activates the categorized resource utilization tracing. It should
1704 be enabled if tracing categories are used by this simulator.
1707 tracing/uncategorized
1709 It activates the uncategorized resource utilization tracing. Use it if
1710 this simulator do not use tracing categories and resource use have to be
1716 A file with this name will be created to register the simulation. The file
1717 is in the Paje format and can be analyzed using Triva or Paje visualization
1718 tools. More information can be found in these webpages:
1719 <a href="http://triva.gforge.inria.fr/">http://triva.gforge.inria.fr/</a>
1720 <a href="http://paje.sourceforge.net/">http://paje.sourceforge.net/</a>
1725 This option only has effect if this simulator is SMPI-based. Traces the MPI
1726 interface and generates a trace that can be analyzed using Gantt-like
1727 visualizations. Every MPI function (implemented by SMPI) is transformed in a
1728 state, and point-to-point communications can be analyzed with arrows.
1733 This option only has effect if this simulator is SMPI-based. The processes
1734 are grouped by the hosts where they were executed.
1739 This option only has effect if this simulator is MSG-based. It traces the
1740 behavior of all categorized MSG tasks, grouping them by hosts.
1745 This option only has effect if this simulator is MSG-based. It traces the
1746 behavior of all categorized MSG processes, grouping them by hosts. This option
1747 can be used to track process location if this simulator has process migration.
1751 triva/categorized:graph_categorized.plist
1753 This option generates a graph configuration file for Triva considering
1754 categorized resource utilization.
1757 triva/uncategorized:graph_uncategorized.plist
1759 This option generates a graph configuration file for Triva considering
1760 uncategorized resource utilization.
1762 \subsubsection faq_tracing_example Example of Instrumentation
1764 A simplified example using the tracing mandatory functions.
1767 int main (int argc, char **argv)
1769 MSG_global_init (&argc, &argv);
1771 //(... after deployment ...)
1773 //note that category declaration must be called after MSG_create_environment
1774 TRACE_category_with_color ("request", "1 0 0");
1775 TRACE_category_with_color ("computation", "0.3 1 0.4");
1776 TRACE_category ("finalize");
1778 m_task_t req1 = MSG_task_create("1st_request_task", 10, 10, NULL);
1779 m_task_t req2 = MSG_task_create("2nd_request_task", 10, 10, NULL);
1780 m_task_t req3 = MSG_task_create("3rd_request_task", 10, 10, NULL);
1781 m_task_t req4 = MSG_task_create("4th_request_task", 10, 10, NULL);
1782 TRACE_msg_set_task_category (req1, "request");
1783 TRACE_msg_set_task_category (req2, "request");
1784 TRACE_msg_set_task_category (req3, "request");
1785 TRACE_msg_set_task_category (req4, "request");
1787 m_task_t comp = MSG_task_create ("comp_task", 100, 100, NULL);
1788 TRACE_msg_set_task_category (comp, "computation");
1790 m_task_t finalize = MSG_task_create ("finalize", 0, 0, NULL);
1791 TRACE_msg_set_task_category (finalize, "finalize");
1800 \subsubsection faq_tracing_analyzing Analyzing the SimGrid Traces
1802 The SimGrid library, during an instrumented simulation, creates a trace file in
1803 the Paje file format that contains the platform utilization for the simulation
1804 that was executed. The visualization analysis of this file is performed with the
1805 visualization tool <a href="http://triva.gforge.inria.fr">Triva</a>, with
1806 special configurations tunned to SimGrid needs. This part of the documentation
1807 explains how to configure and use Triva to analyse a SimGrid trace file.
1809 - <b>Installing Triva</b>: the tool is available in the INRIAGforge,
1810 at <a href="http://triva.gforge.inria.fr">http://triva.gforge.inria.fr</a>.
1811 Use the following command to get the sources, and then check the file
1812 <i>INSTALL</i>. This file contains instructions to install
1813 the tool's dependencies in a Ubuntu/Debian Linux. The tool can also
1814 be compiled in MacOSes natively, check <i>INSTALL.mac</i> file.
1816 $ svn checkout svn://scm.gforge.inria.fr/svn/triva
1821 - <b>Executing Triva</b>: a binary called <i>Triva</i> is available after the
1822 installation (you can execute it passing <em>--help</em> to check its
1823 options). If the triva binary is not available after following the
1824 installation instructions, you may want to execute the following command to
1825 initialize the GNUstep environment variables. We strongly recommend that you
1826 use the latest GNUstep packages, and not the packages available through apt-get
1827 in Ubuntu/Debian packaging systems. If you install GNUstep using the latest
1828 available packages, you can execute this command:
1830 $ source /usr/GNUstep/System/Library/Makefiles/GNUstep.sh
1832 You should be able to see this output after the installation of triva:
1834 $ ./Triva.app/Triva --help
1835 Usage: Triva [OPTIONS...] TRACE0 [TRACE1]
1836 Trace Analysis through Visualization
1839 --ti_frequency {double} Animation: frequency of updates
1840 --ti_hide Hide the TimeInterval window
1841 --ti_forward {double} Animation: value to move time-slice
1842 --ti_apply Apply the configuration
1843 --ti_update Update on slider change
1844 --ti_animate Start animation
1845 --ti_start {double} Start of time slice
1846 --ti_size {double} Size of time slice
1848 --comparison Compare Trace Files (Experimental)
1849 --graph Configurable Graph
1850 --list Print Trace Type Hierarchy
1851 --hierarchy Export Trace Type Hierarchy (dot)
1852 --stat Trace Statistics and Memory Utilization
1853 --instances List All Trace Entities
1854 --linkview Link View (Experimental)
1855 --treemap Squarified Treemap
1856 --merge Merge Trace Files (Experimental)
1857 --check Check Trace File Integrity
1859 --gc_conf {file} Graph Configuration in Property List Format
1860 --gc_apply Apply the configuration
1861 --gc_hide Hide the GraphConfiguration window
1863 Triva expects that the user choose one of the available options
1864 (currently <em>--graph</em> or <em>--treemap</em> for a visualization analysis)
1865 and the trace file from the simulation.
1867 - <b>Understanding Triva - time-slice</b>: the analysis of a trace file using
1868 the tool always takes into account the concept of the <em>time-slice</em>.
1869 This concept means that what is being visualized in the screen is always
1870 calculated considering a specific time frame, with its beggining and end
1871 timestamp. The time-slice is configured by the user and can be changed
1872 dynamically through the window called <em>Time Interval</em> that is opened
1873 whenever a trace file is being analyzed. The next figure depicts the time-slice
1874 configuration window.
1875 In the top of the window, in the space named <i>Trace Time</i>,
1876 the two fields show the beggining of the trace (which usually starts in 0) and
1877 the end (that depends on the time simulated by SimGrid). The middle of the
1878 window, in the square named <i>Time Slice Configuration</i>, contains the
1879 aspects related to the time-slice, including its <i>start</i> and its
1880 <i>size</i>. The gray rectangle in the bottom of this part indicates the
1881 <i>current time-slice</i> that is considered for the drawings. If the checkbox
1882 <i>Update Drawings on Sliders Change</i> is not selected, the button
1883 <i>Apply</i> must be clicked in order to inform triva that the
1884 new time-slice must be considered. The bottom part of the window, in the space
1885 indicated by the square <i>Time Slice Animation</i> can be used to advance
1886 the time-frame automatically. The user configures the amount of time that the
1887 time-frame will forward and how frequent this update will happen. Once this is
1888 configured, the user clicks the <i>Play</i> button in order to see the dynamic
1889 changes on the drawings.
1892 <a href="triva-time_interval.png" border=0><img src="triva-time_interval.png" width="50%" border=0></a>
1895 <b>Remarks:</b> when the trace has too many hosts or links, the computation to
1896 take into account a new time-slice can be expensive. When this happens, the
1897 <i>Frequency</i> parameter, but also updates caused by change on configurations
1898 when the checkbox <i>Update Drawings on Sliders
1899 Change</i> is selected will not be followed.
1901 - <b>Understanding Triva - graph</b>: this part of the documention explains how
1902 to analyze the traces using the graph view of Triva, when the user executes
1903 the tool passing <em>--graph</em> as parameter. Triva opens three windows when
1904 this parameter is used: the <i>Time Interval</i> window (previously described),
1905 the <i>Graph Representation</i> window, and the <em>Graph Configuration</em>
1906 window. The Graph Representation is the window where drawings take place.
1907 Initially, it is completely white waiting for a proper graph configuration input
1908 by the user. We start the description of this type of analysis by describing the
1909 <i>Graph Configuration</i> window (depicted below). By using a particular
1910 configuration, triva
1911 can be used to customize the graph drawing according to
1912 the SimGrid trace that was created with user-specific categories. Before delving
1913 into the details of this customization, let us first explain the major parts of
1914 the graph configuration window. The buttons located in the top-right corner can
1915 be used to delete, copy and create a new configuration. The checkbox in the
1916 top-middle part of the window indicates if the configuration typed in the
1917 textfield is syntactically correct (we are using the non-XML
1918 <a href="http://en.wikipedia.org/wiki/Property_list">Property List Format</a> to
1919 describe the configuration). The pop-up button located on the top-left corner
1920 indicates the selected configuration (the user can have multiple graph
1921 configurations). The bottom-left text field contains the name of the current
1922 configuration (updates on this field must be followed by typing enter on the
1923 keyboard to take into account the name change). The bottom-right <em>Apply</em>
1924 button activates the current configuration, resulting on an update on the graph
1928 <a href="triva-graph_configuration.png" border=0><img src="triva-graph_configuration.png" width="50%" border=0></a>
1931 <b>Basic SimGrid Configuration</b>: The figure shows in the big textfield the
1932 basic configuration that should be used during the analysis of a SimGrid trace
1933 file. The basic logic of the configuration is as follows:
1939 The nodes of the graph will be created based on the <i>node</i> parameter, which
1940 in this case is the different <em>"HOST"</em>s of the platform
1941 used to simulate. The <i>edge</i> parameter indicates that the edges of the
1942 graph will be created based on the <em>"LINK"</em>s of the platform. After the
1943 definition of these two parameters, the configuration must detail how
1944 <em>HOST</em>s and <em>LINK</em>s should be drawn. For that, the configuration
1945 must have an entry for each of the types used. For <em>HOST</em>, as basic
1946 configuration, we have:
1953 The parameter <em>size</em> indicates which variable from the trace file will be
1954 used to define the size of the node HOST in the visualization. If the simulation
1955 was executed with availability traces, the size of the nodes will be changed
1956 according to these traces. The parameter <em>scale</em> indicates if the value
1957 of the variable is <em>global</em> or <em>local</em>. If it is global, the value
1958 will be relative to the power of all other hosts, if it is local, the value will
1959 be relative locally.
1960 For <em>LINK</em> we have:
1970 For the types specified in the <em>edge</em> parameter (such as <em>LINK</em>),
1971 the configuration must contain two additional parameters: <em>src</em> and
1972 <em>dst</em> that are used to properly identify which nodes this edge is
1973 connecting. The values <em>source</em> and <em>destination</em> are always present
1974 in the SimGrid trace file and should not be changed in the configuration. The
1975 parameter <em>size</em> for the LINK, in this case, is configured as the
1976 variable <em>bandwidth</em>, with a <em>global</em> scale. The scale meaning
1977 here is exactly the same used for nodes. The last parameter is the GraphViz
1978 algorithm used to calculate the position of the nodes in the graph
1981 graphviz-algorithm = neato;
1984 <b>Customizing the Graph Representation</b>: triva is capable to handle
1985 a customized graph representation based on the variables present in the trace
1986 file. In the case of SimGrid, every time a category is created for tasks, two
1987 variables in the trace file are defined: one to indicate node utilization (how
1988 much power was used by that task category), and another to indicate link
1989 utilization (how much bandwidth was used by that category). For instance, if the
1990 user declares a category named <i>request</i>, there will be variables named
1991 <b>p</b><i>request</i> and a <b>b</b><i>request</i> (<b>p</b> for power and
1992 <b>b</b> for bandwidth). It is important to notice that the variable
1993 <i>prequest</i> in this case is only available for HOST, and
1994 <i>brequest</i> is only available for LINK. <b>Example</b>: suppose there are
1995 two categories for tasks: request and compute. To create a customized graph
1996 representation with a proportional separation of host and link utilization, use
1997 as configuration for HOST and LINK this:
2006 values = (prequest, pcomputation);
2019 values = (brequest, bcomputation);
2023 Where <i>sep_host</i> contains a composition of type <i>separation</i> where
2024 its max size is the <i>power</i> of the host and the variables <i>prequest</i>
2025 and <i>pcomputation</i> are drawn proportionally to the size of the HOST. And
2026 <i>sep_link</i> is also a separation where max is defined as the
2027 <i>bandwidth</i> of the link, and the variables <i>brequest</i> and
2028 <i>bcomputation</i> are drawn proportionally within a LINK.
2029 <i>This configuration enables the analysis of resource utilization by MSG tasks,
2030 and the identification of load-balancing issues, network bottlenecks, for
2032 <b>Other compositions</b>: besides <i>separation</i>, it is possible to use
2033 other types of compositions, such as gradients, and colors, like this:
2038 values = (numberOfTasks);
2042 values = (is_server);
2045 Where <i>gra_host</i> creates a gradient within a node of the graph, using a
2046 global scale and using as value a variable called <i>numberOfTasks</i>, that
2047 could be declared by the user using the optional tracing functions of SimGrid.
2048 If scale is global, the max and min value for the gradient will be equal to the
2049 max and min numberOfTasks among all hosts, and if scale is local, the max and
2050 min value based on the value of numberOfTasks locally in each host.
2051 And <i>color_host</i> composition draws a square based on a positive value of
2052 the variable <i>is_server</i>, that could also be defined by the user using the
2053 SimGrid tracing functions. \n
2054 <b>The Graph Visualization</b>: The next figure shows a graph visualization of a
2055 given time-slice of the masterslave_forwarder example (present in the SimGrid
2056 sources). The red color indicates tasks from the <i>compute</i> category. This
2057 visualization was generated with the following configuration:
2070 values = (pcompute, pfinalize);
2082 values = (bcompute, bfinalize);
2085 graphviz-algorithm = neato;
2090 <a href="triva-graph_visualization.png" border=0><img src="triva-graph_visualization.png" width="50%" border=0></a>
2094 - <b>Understading Triva - colors</b>: An important issue when using Triva is how
2095 to define colors. To do that, we have to know which variables are defined in
2096 the trace file generated by the SimGrid library. The parameter <em>--list</em>
2097 lists the variables for a given trace file:
2099 $ Triva -l masterslave_forwarder.trace
2117 We can see that HOST has seven variables (from power to pfinalize) and LINK has
2118 four (from bandwidth to bfinalize). To define a red color for the
2119 <i>pcompute</i> and <i>bcompute</i> (which are defined based on user category
2120 <i>compute</i>), execute:
2122 $ defaults write Triva 'pcompute Color' '1 0 0'
2123 $ defaults write Triva 'bcompute Color' '1 0 0'
2125 Where the three numbers in each line are the RGB color with values from 0 to 1.
2127 \subsection faq_modelchecking Model-Checking
2128 \subsubsection faq_modelchecking_howto How to use it
2129 To enable the experimental SimGrid model-checking support the program should
2130 be executed with the command line argument
2134 Properties are expressed as assertions using the function
2136 void MC_assert(int prop);
2139 \subsection faq_binding_lua Lua Binding
2140 Most of Simgrid modules require a good level in C programming, since simgrid is used to be as standard C library.
2141 Sometime users prefer using some kind of « easy scripts » or a language easier to code with, for their works,
2142 which avoid dealing with C errors, and sometime an important gain of time.
2143 Besides Java Binding, Lua and Ruby bindings are available since version 3.4 of Simgrid
2144 for MSG Module, and we are currenlty working on bindings for other modules.
2147 \subsubsection faq_binding_lua_about What is lua ?
2148 Lua is a lightweight, reflective, imperative and functional programming language,
2149 designed as a scripting language with extensible semantics as a primary goal (see official web site <a href="http://www.lua.org">here</a>).
2150 \subsubsection faq_binding_lua_why Why lua ?
2151 Lua is a fast, portable and powerful script language, quite simple to use for developpers.
2152 it combines procedural features with powerful data description facilities,
2153 by using a simple, yet powerful, mechanism of tables.
2154 Lua has a relatively simple C API compared to other scripting languages,
2155 and accordingly it provides a robust, easy to use it.
2156 \subsubsection faq_binding_lua_simgrid How to use lua in Simgrid ?
2157 Actually, the use of lua in Simgrid is quite simple, you have just to follow the same steps as coding with C in Simgird :
2158 - Coding functions coresponding to each process
2159 - loading the platforme/deployment XML file that describe the environment of simulation
2160 - and … Running the Simulation.
2162 \dontinclude lua/masterslave/master.lua
2163 \subsubsection faq_binding_lua_example_master_slave Master/Slave Example
2166 \until end_of_master
2167 we mainly use simgrid.Task.new(task_name,computation_size,communication_size) to create our MSG Task,
2168 then simgrid.Task.send(task,alias) to send it.
2169 we use also simgrid.Task.name(task), to get the task's name.
2171 \dontinclude lua/masterslave/slave.lua
2174 Here, we see the use of simgrid.Task.recv(alias) to receive a task with a specific alias,
2175 this function return directly the task recevied.
2177 \dontinclude lua/masterslave/master_slave.lua
2178 \li Set Environmenet and run application
2179 \until simgrid.clean()
2181 \subsubsection faq_binding_lua_example_data Exchanging Data
2182 You can also exchange data between Process using lua. for that, you have to deal with lua task as a table,
2183 since lua is based itself on a mechanism of tables,
2184 so you can exchange any kind of data (tables, matrix, strings,…) between process via tasks.
2188 task = simgrid.Task.new("data_task",task_comp,task_comm);
2189 task['matrix'] = my_matrix;
2190 task['table'] = my_table;
2191 task['message'] = "Hello from (Lua || Simgrid ) !! "
2193 simgrid.Task.send(task,alias)
2195 After creating task, we associate to it various kind of data with a specific key (string in this case)
2196 to distinguish between data variables. The receiver will use this key to access easily to datas.
2199 \li Receiver processe
2201 task = simgrid.Task.recv(alias);
2202 sender_matrix = task['matrix'];
2203 sender_table = task['table'];
2204 sender_message = task['message']
2207 Note that in lua, both sender and receiver share the same lua task.
2208 So that the receiver could joint data directly on the received task without sending it back.
2209 You can find a complet example (matrix multiplication case) in the file example/lua/mult_matrix.lua.
2212 \subsubsection faq_binding_lua_example_bypass Bypass XML
2213 maybe you wonder if there is a way to bypass the XML files,
2214 and describe your platform directly from the code, with lua bindings it's Possible !! how ?
2215 We provide some additional (tricky?) functions in lua that allows you to set up your own platform without using the XML files
2216 ( this can be useful for large platforms, so a simple for loop will avoid you to deal with an annoying XML File ;) )
2219 \li set Routing mode
2221 simgrid.AS.new{id="AS0",mode="Full"};
2226 simgrid.Host.new{id="Tremblay",power=98095000};
2227 simgrid.Host.new{id="Jupiter",power=76296000};
2228 simgrid.Host.new{id="Fafard",power=76296000};
2229 simgrid.Host.new{id="Ginette",power=48492000};
2230 simgrid.Host.new{id="Bourassa",power=48492000};
2232 we use simgrid.Host.new{id=id_host,power=power_host} to instanciate our hosts.
2237 simgrid.Link.new{id=i,bandwidth=252750+ i*768,latency=0.000270544+i*0.087}; -- some crazy values ;)
2240 we used simgrid.Link.new{id=link_id,bandwidth=bw,latency=lat} with a simple for loop to create all links we need (much easier than XML hein ?)
2244 -- simgrid.Route.new(src_id,des_id,links_nb,links_list)
2245 simgrid.Route.new("Tremblay","Jupiter",1,{"1"});
2246 simgrid.Route.new("Tremblay","Fafard",6,{"0","1","2","3","4","8"});
2247 simgrid.Route.new("Tremblay","Ginette",3,{"3","4","5"});
2248 simgrid.Route.new("Tremblay","Bourassa",7,{"0","1","3","2","4","6","7"});
2250 simgrid.Route.new("Jupiter","Tremblay",1,{"1"});
2251 simgrid.Route.new("Jupiter","Fafard",7,{"0","1","2","3","4","8","9"});
2252 simgrid.Route.new("Jupiter","Ginette",4,{"3","4","5","9"});
2253 simgrid.Route.new("Jupiter","Bourassa",8,{"0","1","2","3","4","6","7","9"});
2256 for each host you have to specify which route to choose to access to the rest of hosts connected in the grid.
2260 simgrid.register_platform();
2262 Don't forget to register your platform, that SURF callbacks starts their work ;)
2266 simgrid.Host.setFunction("Tremblay","Master",4,{"20","550000000","1000000","4"});
2267 simgrid.Host.setFunction("Bourassa","Slave",1,{"0"});
2268 simgrid.Host.setFunction("Jupiter","Slave",1,{"1"});
2269 simgrid.Host.setFunction("Fafard","Slave",1,{"2"});
2270 simgrid.Host.setFunction("Ginette","Slave",1,{"3"});
2272 you don't need to use a deployment XML file, thanks to simgrid.Host.setFunction(host_id,function,args_number,args_list)
2273 you can associate functions for each host with arguments if needed .
2277 simgrid.register_application();
2279 Yes, Here too you have to resgiter your application before running the simulation.
2281 the full example is distributed in the file examples/lua/master_slave_bypass.lua
2283 \subsection faq_binding_ruby Ruby Binding
2286 \subsubsection faq_binding_ruby_simgrid Use Ruby in Simgrid
2287 Since v3.4, the use of <a href="http://ruby-lang.org">ruby</a> in simgrid is available for the MSG Module.
2288 you can find almost all MSG functionalities in Ruby code, that allows you to set up your environment, manage tasks between hosts and run the simulation.
2290 \subsubsection faq_binding_ruby_example Master/Slave Ruby Application
2291 for each process method(master and slave in this example), you have to associate a ruby class, that should inherit from <i>MSG::Process</i> ruby class,
2292 with a 'main' function that describe the behaviour of the process during the simulation.
2301 class Master < MSG::Process
2302 # main : that function that will be executed when running simulation
2304 def main(args) # args is an array containing arguments for function master
2307 MSG::info("args["+String(i)+"]="+args[i])
2310 raise "Master needs 3 arguments" if size < 3
2311 numberOfTask = Integer(args[0])
2312 taskComputeSize = Float(args[1])
2313 taskCommunicationSize = Float(args[2])
2314 slaveCount = Integer(args[3])
2316 # Creates and sends the tasks
2317 for i in 0..numberOfTask-1
2318 task = Task.new("Task_"+ i.to_s, taskComputeSize , taskCommunicationSize);
2319 mailbox = "slave " + (i%slaveCount).to_s
2320 MSG::info("Master Sending "+ task.name + " to " + mailbox + " with Comput Size " +
2323 MSG::info("Master Done Sending " + task.name + " to " + mailbox)
2326 # Sending Finalize MSG::Tasks
2327 MSG::info("Master: All tasks have been dispatched. Let's tell everybody the computation is over.")
2328 for i in 0..slaveCount-1
2329 mailbox = "slave " + i.to_s
2330 finalize_task = Task.new("finalize",0,0)
2331 finalize_task.send(mailbox)
2333 MSG::info("Master : Everything's Done")
2339 the class MSG::Task contains methods that allows the management of the native MSG tasks.
2340 in master ruby code we used :
2341 - <i>MSG::Task.new(task_name,compute_size,communication_size)</i> : to instanciate a new task.
2342 - <i>MSG::Task.send(mailbox)</i> : to send the task via a mailbox alias.
2343 - <i>MSG::Task.name</i> : to get the task's name.
2347 class Slave < MSG::Process
2350 mailbox = "slave " + args[0]
2351 for i in 0..args.size-1
2352 MSG::debug("args["+String(i)+"]="+args[i])
2356 MSG::info("Slave '"+ mailbox +"' waiting for new task");
2357 task = Task.receive(mailbox)
2358 if (task.name == "finalize")
2362 MSG::info("Slave '" + mailbox + "' done executing task "+ task.name + ".")
2364 MSG::info("I'm done, see you")
2368 to receive a task, we use the method <i>MSG::Task.receive(mailbox)</i> that return a MSG:Task object (received task).
2377 if (ARGV.length == 2)
2378 MSG.createEnvironment(ARGV[0])
2379 MSG.deployApplication(ARGV[1])
2383 MSG.createEnvironment("platform.xml")
2384 MSG.deployApplication("deploy.xml")
2387 puts "Simulation time : " + MSG.getClock.to_s
2391 - <i>MSG.createEnvironment(platform_file)</i> : set up the environment
2392 - <i>MSG.deployApplication(deployment_file)</i> : load the deployment file description.
2393 - <i>MSG.run</i> : run the simulation
2395 \subsubsection faq_binding_ruby_data Exchanging data
2396 ruby bindings provides two ways to exchange data between ruby processes.
2397 \li MSG::Task.join & MSG::Task.data <br/>
2399 the MSG::Task class contains 2 methods that allows a data exchange between 2 process.
2401 -<i>MSG::Task.join</i> : makes possible to join any kind of ruby data within a task.
2405 myTable <<1<<-2<<45<<67<<87<<76<<89<<56<<78<<3<<-4<<99
2406 # Creates and send Task With the Table inside
2407 task = MSG::Task.new("quicksort_task",taskComputeSize, taskCommunicationSize);
2412 -<i>MSG::Task.data</i> : to access to the data contained into the task.
2415 task = MSG::Task.receive(recv_mailbox.to_s)
2417 quicksort(table,0,table.size-1)
2420 you can find a complet example illustrating the use of those methods in file /example/ruby/Quicksort.rb
2424 another 'object-oriented' way to do it, is to make your own 'task' class that inherit from MSG::Task ,
2425 and contains data you want to deal with, the only 'tricky' thing is that "the initializer" method has no effect !
2427 the use of some getter/setter methods would be the simple way to manage your data :)
2429 class PingPongTask < MSG::Task
2430 # The initialize method has no effect
2440 you can find an example of use in file example/ruby/PingPong.rb
2442 \section faq_troubleshooting Troubleshooting
2444 \subsection faq_trouble_lib_compil SimGrid compilation and installation problems
2446 \subsubsection faq_trouble_lib_config cmake fails!
2448 We know only one reason for the configure to fail:
2450 - <b>You are using a broken build environment</b>\n
2451 If symptom is that the configury magic complains about gcc not being able to build
2452 executables, you are probably missing the libc6-dev package. Damn Ubuntu.
2454 If you experience other kind of issue, please get in touch with us. We are
2455 always interested in improving our portability to new systems.
2457 \subsubsection faq_trouble_distcheck Dude! "ctest" fails on my machine!
2459 Don't assume we never run this target, because we do. Check
2460 http://cdash.inria.fr/CDash/index.php?project=Simgrid (click on
2461 previous if there is no result for today: results are produced only by
2462 11am, French time) and
2463 https://buildd.debian.org/status/logs.php?pkg=simgrid if you don't believe us.
2465 If it's failing on your machine in a way not experienced by the
2466 autobuilders above, please drop us a mail on the mailing list so that
2467 we can check it out. Make sure to read \ref faq_bugrepport before you
2470 \subsection faq_trouble_compil User code compilation problems
2472 \subsubsection faq_trouble_err_logcat "gcc: _simgrid_this_log_category_does_not_exist__??? undeclared (first use in this function)"
2474 This is because you are using the log mecanism, but you didn't created
2475 any default category in this file. You should refer to \ref XBT_log
2476 for all the details, but you simply forgot to call one of
2477 XBT_LOG_NEW_DEFAULT_CATEGORY() or XBT_LOG_NEW_DEFAULT_SUBCATEGORY().
2479 \subsubsection faq_trouble_pthreadstatic "gcc: undefined reference to pthread_key_create"
2481 This indicates that one of the library SimGrid depends on (libpthread
2482 here) was missing on the linking command line. Dependencies of
2483 libsimgrid are expressed directly in the dynamic library, so it's
2484 quite impossible that you see this message when doing dynamic linking.
2486 If you compile your code statically (and if you use a pthread version
2487 of SimGrid -- see \ref faq_more_processes), you must absolutely
2488 specify <tt>-lpthread</tt> on the linker command line. As usual, this should
2489 come after <tt>-lsimgrid</tt> on this command line.
2491 \subsection faq_trouble_errors Runtime error messages
2493 \subsubsection faq_flexml_limit "surf_parse_lex: Assertion `next limit' failed."
2495 This is because your platform file is too big for the parser.
2497 Actually, the message comes directly from FleXML, the technology on top of
2498 which the parser is built. FleXML has the bad idea of fetching the whole
2499 document in memory before parsing it. And moreover, the memory buffer size
2500 must be determined at compilation time.
2502 We use a value which seems big enough for our need without bloating the
2503 simulators footprints. But of course your mileage may vary. In this case,
2504 just edit src/surf/surfxml.l modify the definition of
2505 FLEXML_BUFFERSTACKSIZE. E.g.
2508 #define FLEXML_BUFFERSTACKSIZE 1000000000
2511 Then recompile and everything should be fine, provided that your version of
2512 Flex is recent enough (>= 2.5.31). If not the compilation process should
2515 A while ago, we worked on FleXML to reduce a bit its memory consumption, but
2516 these issues remain. There is two things we should do:
2518 - use a dynamic buffer instead of a static one so that the only limit
2519 becomes your memory, not a stupid constant fixed at compilation time
2520 (maybe not so difficult).
2521 - change the parser so that it does not need to get the whole file in
2522 memory before parsing
2523 (seems quite difficult, but I'm a complete newbe wrt flex stuff).
2525 These are changes to FleXML itself, not SimGrid. But since we kinda hijacked
2526 the development of FleXML, I can grant you that any patches would be really
2527 welcome and quickly integrated.
2529 <b>Update:</b> A new version of FleXML (1.7) was released. Most of the work
2530 was done by William Dowling, who use it in his own work. The good point is
2531 that it now use a dynamic buffer, and that the memory usage was greatly
2532 improved. The downside is that William also changed some things internally,
2533 and it breaks the hack we devised to bypass the parser, as explained in
2534 \ref faq_flexml_bypassing. Indeed, this is not a classical usage of the
2535 parser, and Will didn't imagine that we may have used (and even documented)
2536 such a crude usage of FleXML. So, we now have to repair the bypassing
2537 functionality to use the lastest FleXML version and fix the memory usage in
2540 \subsubsection faq_trouble_gras_transport GRAS spits networking error messages
2542 Gras, on real platforms, naturally use regular sockets to communicate. They
2543 are deeply hidden in the gras abstraction, but when things go wrong, you may
2544 get some weird error messages. Here are some example, with the probable
2547 - <b>Transport endpoint is not connected</b>: several processes try to open
2548 a server socket on the same port number of the same machine. This is
2549 naturally bad and each process should pick its own port number for this.\n
2550 Maybe, you just have some processes remaining from a previous experiment
2552 Killing them may help, but again if you kill -KILL them, you'll have to
2553 wait for a while: they didn't close there sockets properly and the system
2554 needs a while to notice that this port is free again.
2556 - <b>Socket closed by remote side</b>: if the remote process is not
2557 supposed to close the socket at this point, it may be dead.
2559 - <b>Connection reset by peer</b>: I found this on Internet about this
2560 error. I think it's what's happening here, too:\n
2561 <i>This basically means that a network error occurred while the client was
2562 receiving data from the server. But what is really happening is that the
2563 server actually accepts the connection, processes the request, and sends
2564 a reply to the client. However, when the server closes the socket, the
2565 client believes that the connection has been terminated abnormally
2566 because the socket implementation sends a TCP reset segment telling the
2567 client to throw away the data and report an error.\n
2568 Sometimes, this problem is caused by not properly closing the
2569 input/output streams and the socket connection. Make sure you close the
2570 input/output streams and socket connection properly. If everything is
2571 closed properly, however, and the problem persists, you can work around
2572 it by adding a one-second sleep before closing the streams and the
2573 socket. This technique, however, is not reliable and may not work on all
2575 Since GRAS sockets are closed properly (repeat after me: there is no bug
2576 in GRAS), it is either that you are closing your sockets on server side
2577 before the client get a chance to read them (use gras_os_sleep() to delay
2578 the server), or the server died awfully before the client got the data.
2580 \subsubsection faq_trouble_errors_big_fat_warning I'm told that my XML files are too old.
2582 The format of the XML platform description files is sometimes
2583 improved. For example, we decided to change the units used in SimGrid
2584 from MBytes, MFlops and seconds to Bytes, Flops and seconds to ease
2585 people exchanging small messages. We also reworked the route
2586 descriptions to allow more compact descriptions.
2588 That is why the XML files are versionned using the 'version' attribute
2589 of the root tag. Currently, it should read:
2591 <platform version="2">
2594 If your files are too old, you can use the simgrid_update_xml.pl
2595 script which can be found in the tools directory of the archive.
2597 \subsection faq_trouble_valgrind Valgrind-related and other debugger issues
2599 If you don't, you really should use valgrind to debug your code, it's
2602 \subsubsection faq_trouble_vg_longjmp longjmp madness in valgrind
2604 This is when valgrind starts complaining about longjmp things, just like:
2606 \verbatim ==21434== Conditional jump or move depends on uninitialised value(s)
2607 ==21434== at 0x420DBE5: longjmp (longjmp.c:33)
2609 ==21434== Use of uninitialised value of size 4
2610 ==21434== at 0x420DC3A: __longjmp (__longjmp.S:48)
2613 This is the sign that you didn't used the exception mecanism well. Most
2614 probably, you have a <tt>return;</tt> somewhere within a <tt>TRY{}</tt>
2615 block. This is <b>evil</b>, and you must not do this. Did you read the section
2618 \subsubsection faq_trouble_vg_libc Valgrind spits tons of errors about backtraces!
2620 It may happen that valgrind, the memory debugger beloved by any decent C
2621 programmer, spits tons of warnings like the following :
2622 \verbatim ==8414== Conditional jump or move depends on uninitialised value(s)
2623 ==8414== at 0x400882D: (within /lib/ld-2.3.6.so)
2624 ==8414== by 0x414EDE9: (within /lib/tls/i686/cmov/libc-2.3.6.so)
2625 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
2626 ==8414== by 0x414F937: _dl_open (in /lib/tls/i686/cmov/libc-2.3.6.so)
2627 ==8414== by 0x4150F4C: (within /lib/tls/i686/cmov/libc-2.3.6.so)
2628 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
2629 ==8414== by 0x415102D: __libc_dlopen_mode (in /lib/tls/i686/cmov/libc-2.3.6.so)
2630 ==8414== by 0x412D6B9: backtrace (in /lib/tls/i686/cmov/libc-2.3.6.so)
2631 ==8414== by 0x8076446: xbt_dictelm_get_ext (dict_elm.c:714)
2632 ==8414== by 0x80764C1: xbt_dictelm_get (dict_elm.c:732)
2633 ==8414== by 0x8079010: xbt_cfg_register (config.c:208)
2634 ==8414== by 0x806821B: MSG_config (msg_config.c:42)
2637 This problem is somewhere in the libc when using the backtraces and there is
2638 very few things we can do ourselves to fix it. Instead, here is how to tell
2639 valgrind to ignore the error. Add the following to your ~/.valgrind.supp (or
2640 create this file on need). Make sure to change the obj line according to
2641 your personnal mileage (change 2.3.6 to the actual version you are using,
2642 which you can retrieve with a simple "ls /lib/ld*.so").
2645 name: Backtrace madness
2647 obj:/lib/ld-2.3.6.so
2652 fun:__libc_dlopen_mode
2655 Then, you have to specify valgrind to use this suppression file by passing
2656 the <tt>--suppressions=$HOME/.valgrind.supp</tt> option on the command line.
2657 You can also add the following to your ~/.bashrc so that it gets passed
2658 automatically. Actually, it passes a bit more options to valgrind, and this
2659 happen to be my personnal settings. Check the valgrind documentation for
2662 \verbatim export VALGRIND_OPTS="--leak-check=yes --leak-resolution=high --num-callers=40 --tool=memcheck --suppressions=$HOME/.valgrind.supp" \endverbatim
2664 \subsubsection faq_trouble_backtraces Truncated backtraces
2666 When debugging SimGrid, it's easier to pass the
2667 --disable-compiler-optimization flag to the configure if valgrind or
2668 gdb get fooled by the optimization done by the compiler. But you
2669 should remove these flag when everything works before going in
2670 production (before launching your 1252135 experiments), or everything
2671 will run only one half of the true SimGrid potential.
2673 \subsection faq_deadlock There is a deadlock in my code!!!
2675 Unfortunately, we cannot debug every code written in SimGrid. We
2676 furthermore believe that the framework provides ways enough
2677 information to debug such informations yourself. If the textual output
2678 is not enough, Make sure to check the \ref faq_visualization FAQ entry to see
2679 how to get a graphical one.
2681 Now, if you come up with a really simple example that deadlocks and
2682 you're absolutely convinced that it should not, you can ask on the
2683 list. Just be aware that you'll be severely punished if the mistake is
2684 on your side... We have plenty of FAQ entries to redact and new
2685 features to implement for the impenitents! ;)
2687 \subsection faq_surf_network_latency I get weird timings when I play with the latencies.
2689 OK, first of all, remember that units should be Bytes, Flops and
2690 Seconds. If you don't use such units, some SimGrid constants (e.g. the
2691 SG_TCP_CTE_GAMMA constant used in most network models) won't have the
2692 right unit and you'll end up with weird results.
2694 Here is what happens with a single transfer of size L on a link
2695 (bw,lat) when nothing else happens.
2698 0-----lat--------------------------------------------------t
2699 |-----|**** real_bw =min(bw,SG_TCP_CTE_GAMMA/(2*lat)) *****|
2702 In more complex situations, this min is the solution of a complex
2703 max-min linear system. Have a look
2704 <a href="http://lists.gforge.inria.fr/pipermail/simgrid-devel/2006-April/thread.html">here</a>
2705 and read the two threads "Bug in SURF?" and "Surf bug not
2706 fixed?". You'll have a few other examples of such computations. You
2707 can also read "A Network Model for Simulation of Grid Application" by
2708 Henri Casanova and Loris Marchal to have all the details. The fact
2709 that the real_bw is smaller than bw is easy to understand. The fact
2710 that real_bw is smaller than SG_TCP_CTE_GAMMA/(2*lat) is due to the
2711 window-based congestion mechanism of TCP. With TCP, you can't exploit
2712 your huge network capacity if you don't have a good round-trip-time
2713 because of the acks...
2715 Anyway, what you get is t=lat + L/min(bw,SG_TCP_CTE_GAMMA/(2*lat)).
2717 * if I you set (bw,lat)=(100 000 000, 0.00001), you get t = 1.00001 (you fully
2719 * if I you set (bw,lat)=(100 000 000, 0.0001), you get t = 1.0001 (you're on the
2721 * if I you set (bw,lat)=(100 000 000, 0.001), you get t = 10.001 (ouch!)
2723 This bound on the effective bandwidth of a flow is not the only thing
2724 that may make your result be unexpected. For example, two flows
2725 competing on a saturated link receive an amount of bandwidth inversely
2726 proportional to their round trip time.
2728 \subsection faq_bugrepport So I've found a bug in SimGrid. How to report it?
2730 We do our best to make sure to hammer away any bugs of SimGrid, but this is
2731 still an academic project so please be patient if/when you find bugs in it.
2732 If you do, the best solution is to drop an email either on the simgrid-user
2733 or the simgrid-devel mailing list and explain us about the issue. You can
2734 also decide to open a formal bug report using the
2735 <a href="https://gforge.inria.fr/tracker/?atid=165&group_id=12&func=browse">relevant
2736 interface</a>. You need to login on the server to get the ability to submit
2739 We will do our best to solve any problem repported, but you need to help us
2740 finding the issue. Just telling "it segfault" isn't enough. Telling "It
2741 segfaults when running the attached simulator" doesn't really help either.
2742 You may find the following article interesting to see how to repport
2743 informative bug repports:
2744 http://www.chiark.greenend.org.uk/~sgtatham/bugs.html (it is not SimGrid
2745 specific at all, but it's full of good advices).
2747 \author Arnaud Legrand (arnaud.legrand::imag.fr)
2748 \author Martin Quinson (martin.quinson::loria.fr)
2753 ******************************************************************
2754 * OLD CRUFT NOT USED ANYMORE *
2755 ******************************************************************
2758 \subsection faq_crosscompile Cross-compiling a Windows DLL of SimGrid from linux
2760 At the moment, we do not distribute Windows pre-compiled version of SimGrid
2761 because the support for this platform is still experimental. We know that
2762 some parts of the GRAS environment do not work, and we think that the others
2763 environments (MSG and SD) have good chances to work, but we didn't test
2764 ourselves. This section explains how we generate the SimGrid DLL so that you
2765 can build it for yourself. First of all, you need to have a version more
2766 recent than 3.1 (ie, a SVN version as time of writting).
2768 In order to cross-compile the package to windows from linux, you need to
2769 install mingw32 (minimalist gnu win32). On Debian, you can do so by
2770 installing the packages mingw32 (compiler), mingw32-binutils (linker and
2771 so), mingw32-runtime.
2773 You can use the VPATH support of configure to compile at the same time for
2774 linux and windows without dupplicating the source nor cleaning the tree
2775 between each. Just run bootstrap (if you use the SVN) to run the autotools.
2776 Then, create a linux and a win directories. Then, type:
2777 \verbatim cd linux; ../configure --srcdir=.. <usual configure flags>; make; cd ..
2778 cd win; ../configure --srcdir=.. --host=i586-mingw32msvc <flags>; make; cd ..
2780 The trick to VPATH builds is to call configure from another directory,
2781 passing it an extra --srcdir argument to tell it where all the sources are.
2782 It will understand you want to use VPATH. Then, the trick to cross-compile
2783 is simply to add a --host argument specifying the target you want to build
2784 for. The i586-mingw32msvc string is what you have to pass to use the mingw32
2785 environment as distributed in Debian.
2787 After that, you can run all make targets from both directories, and test
2788 easily that what you change for one arch does not break the other one.
2790 It is possible that this VPATH build thing breaks from time to time in the
2791 SVN since it's quite fragile, but it's granted to work in any released
2792 version. If you experience problems, drop us a mail.
2794 Another possible source of issue is that at the moment, building the
2795 examples request to use the gras_stub_generator tool, which is a compiled
2796 program, not a script. In cross-compilation, you need to cross-execute with
2797 wine for example, which is not really pleasant. We are working on this, but
2798 in the meanwhile, simply don't build the examples in cross-compilation
2799 (<tt>cd src</tt> before running make).
2801 Program (cross-)compiled with mingw32 do request an extra DLL at run-time to be
2802 usable. For example, if you want to test your build with wine, you should do
2803 the following to put this library where wine looks for DLLs.
2805 cp /usr/share/doc/mingw32-runtime/mingwm10.dll.gz ~/.wine/c/windows/system/
2806 gunzip ~/.wine/c/windows/system/mingwm10.dll.gz
2809 The DLL is built in src/.libs, and installed in the <i>prefix</i>/bin directory
2810 when you run make install.
2812 If you want to use it in a native project on windows, you need to use
2813 simgrid.dll and mingwm10.dll. For each DLL, you need to build .def file
2814 under linux (listing the defined symbols), and convert it into a .lib file
2815 under windows (specifying this in a way that windows compilers like). To
2816 generate the def files, run (under linux):
2817 \verbatim echo "LIBRARY libsimgrid-0.dll" > simgrid.def
2818 echo EXPORTS >> simgrid.def
2819 nm libsimgrid-0.dll | grep ' T _' | sed 's/.* T _//' >> simgrid.def
2820 nm libsimgrid-0.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> simgrid.def
2822 echo "LIBRARY mingwm10.dll" > mingwm10.def
2823 echo EXPORTS >> mingwm10.def
2824 nm mingwm10.dll | grep ' T _' | sed 's/.* T _//' >> mingwm10.def
2825 nm mingwm10.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> mingwm10.def
2828 To create the import .lib files, use the <tt>lib</tt> windows tool (from
2829 MSVC) the following way to produce simgrid.lib and mingwm10.lib
2830 \verbatim lib /def:simgrid.def
2831 lib /def:mingwm10.def
2834 If you happen to use Borland C Builder, the right command line is the
2835 following (note that you don't need any file.def to get this working).
2836 \verbatim implib simgrid.lib libsimgrid-0.dll
2837 implib mingwm10.lib mingwm10.dll
2840 Then, set the following parameters in Visual C++ 2005:
2841 Linker -> Input -> Additional dependencies = simgrid.lib mingwm10.lib
2843 Just in case you wonder how to generate a DLL from libtool in another
2844 project, we added -no-undefined to any lib*_la_LDFLAGS variables so that
2845 libtool accepts to generate a dynamic library under windows. Then, to make
2846 it true, we pass any dependencies (such as -lws2 under windows or -lpthread
2847 on need) on the linking line. Passing such deps is a good idea anyway so
2848 that they get noted in the library itself, avoiding the users to know about
2849 our dependencies and put them manually on their compilation line. Then we
2850 added the AC_LIBTOOL_WIN32_DLL macro just before AC_PROG_LIBTOOL in the
2851 configure.ac. It means that we exported any symbols which need to be.
2852 Nowadays, functions get automatically exported, so we don't need to load our
2853 header files with tons of __declspec(dllexport) cruft. We only need to do so
2854 for data, but there is no public data in SimGrid so we are good.
