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>
119 \subsubsection faq_cmakeoption1 Liste of options
122 "cmake -D[name]=[value] ... ./"
124 [name] enable_gtnets [value] ON/OFF or TRUE/FALSE or 1/0
125 enable_lua ON/OFF or TRUE/FALSE or 1/0
126 enable_compile_optimizations ON/OFF or TRUE/FALSE or 1/0
127 enable_compile_warnings ON/OFF or TRUE/FALSE or 1/0
128 enable_smpi ON/OFF or TRUE/FALSE or 1/0
129 enable_maintainer_mode ON/OFF or TRUE/FALSE or 1/0
130 enable_supernovae ON/OFF or TRUE/FALSE or 1/0
131 enable_tracing ON/OFF or TRUE/FALSE or 1/0
132 enable_coverage ON/OFF or TRUE/FALSE or 1/0
133 enable_memcheck ON/OFF or TRUE/FALSE or 1/0
134 enable_model-checking ON/OFF or TRUE/FALSE or 1/0
135 enable_debug ON/OFF or TRUE/FALSE or 1/0
136 enable_jedule ON/OFF or TRUE/FALSE or 1/0
137 enable_latency_bound_tracking ON/OFF or TRUE/FALSE or 1/0
138 enable_lib_static ON/OFF or TRUE/FALSE or 1/0
139 enable_pcre ON/OFF or TRUE/FALSE or 1/0
141 gtnets_path <path_to_gtnets_directory>
142 CMAKE_INSTALL_PREFIX <path_to_install_directory>
143 pipol_user <pipol_username>
146 \subsubsection faq_cmakeoption2 Options explaination
148 \li enable_gtnets : set to true implie that user wants to use gtnets.
150 \li enable_lua : set to true implie that user wants to add lua langage into simgrid compilation.
152 \li enable_compile_optimizations : add flags "-O3 -finline-functions -funroll-loops -fno-strict-aliasing"
154 \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"
156 \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.
158 \li enable_maintainer_mode : set to true it remakes some files.
160 \li enable_supernovae : set to true make one file for each lib and compile with those generated files.
162 \li enable_tracing : To enable the generation of simulation traces for visualization.
164 \li enable_coverage : When set to true this option enable code coverage by setting -fprofile-arcs -ftest-coverage flags.
166 \li enable_memcheck : When set to true this option enable tests for memcheck.
168 \li enable_model-checking : Enable the model checking when set to true.
170 \li enable_debug : If enable_debug is set to 'off' Simgrid compil flag has '-DNDEBUG' option.
172 \li enable_jedule : To enable jedule mode, which creates visualizations of task schedules with Simdag.
174 \li enable_latency_bound_tracking : Set to on if you want to be warned when communications are limited by round trip time.
176 \li enable_lib_static : Enable generated Simgrid and smpi static libraries.
178 \li enable_pcre : Use or not the pcre lib for memory optimization.
180 \li custom_flags : If user wants to use a specific flag during compilation, give here.
182 \li gtnets_path : Path to gtnets install directory (ex /usr)
184 \li CMAKE_INSTALL_PREFIX : Path where are installed lib/ doc/ and include/ directories (ex /usr/local)
186 \li pipol_user : specify your pipol username if you want to use the pipol-remote command.
188 \subsubsection faq_cmakeoption3 Initialisation
190 Those options are initialized the first time you launch "cmake ." whithout specified option.
198 enable_compile_optimizations on
201 enable_compile_warnings off
202 enable_maintainer_mode off
205 enable_model-checking off
207 enable_latency_bound_tracking off
208 enable_lib_static off
209 CMAKE_INSTALL_PREFIX /usr/local
215 \subsubsection faq_cmakeoption4 Option's cache and how to reset?
217 When options have been set they are keep into a cache file named "CMakeCache.txt". So if you want
218 reset values you just delete this file located to the project directory.
220 \subsection faq_cmakecompilation Cmake compilation
222 \subsubsection faq_cmakecompilation1 With command line.
225 cmake -D[name]=[value] ... ./
232 cmake -G"Unix Makefiles" -D[name]=[value] ... ./
236 \subsubsection faq_cmakecompilation2 With ccmake tool.
241 Then follow instructions.
243 \subsubsection faq_cmakecompilation2bis Build out of source.
245 As cmake generate many files used for compilation, we recommand to make a build directory.
246 For examples you can make :
249 "navarrop@caraja:~/Developments$ cd simgrid/"
250 "navarrop@caraja:~/Developments/simgrid$ mkdir build_directory"
251 "navarrop@caraja:~/Developments/simgrid$ cd build_directory/"
252 "navarrop@caraja:~/Developments/simgrid/build_directory$ cmake ../"
253 "navarrop@caraja:~/Developments/simgrid/build_directory$ make"
256 Or complety out of sources :
259 "navarrop@caraja:~/Developments$ mkdir build_dir"
260 "navarrop@caraja:~/Developments$ cd build_dir/"
261 "navarrop@caraja:~/Developments/build_dir$ cmake ../simgrid/"
262 "navarrop@caraja:~/Developments/build_dir$ make"
265 Those two kind of compilation permit to delete files created by compilation easier.
267 \subsubsection faq_cmakecompilation3 Resume of command line
271 cmake <path> configure the project
272 make build all targets
273 make VERBOSE=1 build all targets and print build command lines
274 make check test all targets and summarize
275 make dist make the distrib
276 make distcheck check the dist (make + make dist + make check)
277 make install install the project (doc/ bin/ lib/ include/)
278 make uninstall uninstall the project (doc/ bin/ lib/ include/)
279 make clean clean all targets
280 make simgrid_documentation Create simgrid documentation
283 When the project have been succesfully compiling and build you can make tests.
287 ctest launch only tests
289 ctest -D Continuous(Start|Update|Configure|Build)
290 ctest -D Continuous(Test|Coverage|MemCheck|Submit)
291 ctest -D Experimental
292 ctest -D Experimental(Start|Update|Configure|Build)
293 ctest -D Experimental(Test|Coverage|MemCheck|Submit)
295 ctest -D Nightly(Start|Update|Configure|Build)
296 ctest -D Nightly(Test|Coverage|MemCheck|Submit)
297 ctest -D NightlyMemoryCheck
300 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>.
302 \subsection faq_cmakeinstall How to install with cmake?
304 \subsubsection faq_cmakeinstall1 From svn.
308 cmake -Denable_maintainer_mode=on -DCMAKE_INSTALL_PREFIX=/home/navarrop/Bureau/install_simgrid ./
316 cmake -G"Unix Makefiles" -DCMAKE_INSTALL_PREFIX=C:\simgrid_install ./
321 \subsubsection faq_cmakeinstall2 From a distrib
324 For version 3.4.1 and 3.4
325 cmake -Dprefix=/home/navarrop/Bureau/install_simgrid ./
329 cmake -DCMAKE_INSTALL_PREFIX=/home/navarrop/Bureau/install_simgrid ./
334 \subsection faq_cmakeWHATisInstall What is installed by cmake?
336 \subsubsection faq_cmakeWHATisInstallBIN CMAKE_INSTALL_PREFIX/bin
348 \subsubsection faq_cmakeWHATisInstallDOC CMAKE_INSTALL_PREFIX/doc
353 \subsubsection faq_cmakeWHATisInstallINCLUDE CMAKE_INSTALL_PREFIX/include
369 \subsubsection faq_cmakeWHATisInstallLIB CMAKE_INSTALL_PREFIX/lib
374 libsimgrid.so -> libsimgrid.so.3.5
375 libgras.so -> libgras.so.3.5
376 libsmpi.so -> libsmpi.so.3.5
377 lua/5.1/simgrid.so -> ../../libsimgrid.so
378 ruby/1.9.0/x86_64-linux/libsimgrid.so -> ../../../libsimgrid.so
379 ruby/1.9.0/x86_64-linux/simgrid.rb
381 \subsection faq_cmakehowto How to modified sources files for developers
383 \subsubsection faq_cmakehowto1 Add an executable or examples.
385 If you want make an executable you have to create a CMakeList.txt to the src directory.
386 You must specified where to create the executable, source list, dependencies and the name of the binary.
389 cmake_minimum_required(VERSION 2.6)
391 set(EXECUTABLE_OUTPUT_PATH "./")
392 set(LIBRARY_OUTPUT_PATH "${CMAKE_HOME_DIRECTORY}/lib")
394 add_executable(get_sender get_sender.c) #add_executable(<name_of_target> <src list>)
396 ### Add definitions for compile
397 target_link_libraries(get_sender simgrid m pthread) #target_link_libraries(<name_of_targe> <dependencies>)
400 Then you have to modified <project/directory>/buildtools/Cmake/MakeExeLib.cmake and add
403 add_subdirectory(${CMAKE_HOME_DIRECTORY}/<path_where_is_CMakeList.txt>)
406 \subsubsection faq_cmakehowto2 Delete/add sources to lib.
408 If you want modified, add or delete source files from a library you have to edit <project/directory>/buildtools/Cmake/DefinePackages.cmake
412 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/MsgException.java
413 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/JniException.java
414 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/NativeException.java
415 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/HostNotFoundException.java
416 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/ProcessNotFoundException.java
417 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Msg.java
418 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Process.java
419 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Host.java
420 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Task.java
421 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/MsgNative.java
422 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/ApplicationHandler.java
423 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Sem.java
427 \subsubsection faq_cmakehowto3 Add test
429 If you want modified, add or delete tests you have to edit <project/directory>/buildtools/Cmake/AddTests.cmake
430 with this function : ADD_TEST(<name> <bin> <ARGS>)
433 add_test(test-simdag-1 ${CMAKE_HOME_DIRECTORY}/testsuite/simdag/sd_test --cfg=path:${CMAKE_HOME_DIRECTORY}/testsuite/simdag small_platform_variable.xml)
436 \subsection faq_PIPOL Pipol-remote
438 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 :
440 amd64_kvm-linux-debian-lenny
441 amd64_kvm-linux-debian-testing
443 amd64-linux-centos-5.dd.gz
444 amd64-linux-debian-etch.dd.gz
445 amd64-linux-debian-lenny.dd.gz
446 amd64-linux-debian-testing.dd.gz
447 amd64-linux-fedora-core10.dd.gz
448 amd64-linux-fedora-core11.dd.gz
449 amd64-linux-fedora-core12.dd.gz
450 amd64-linux-fedora-core13.dd.gz
451 amd64-linux-fedora-core7.dd.gz
452 amd64-linux-fedora-core8.dd.gz
453 amd64-linux-fedora-core9.dd.gz
454 amd64-linux-mandriva-2007_springs_powerpack.dd.gz
455 amd64-linux-mandriva-2009_powerpack.dd.gz
456 amd64-linux-opensuse-11.dd.gz
457 amd64-linux-redhatEL-5.0.dd.gz
458 amd64-linux-suse-LES10.dd.gz
459 amd64-linux-ubuntu-feisty.dd.gz
460 amd64-linux-ubuntu-hardy.dd.gz
461 amd64-linux-ubuntu-intrepid.dd.gz
462 amd64-linux-ubuntu-jaunty.dd.gz
463 amd64-linux-ubuntu-karmic.dd.gz
464 amd64-linux-ubuntu-lucid.dd.gz
465 amd64-unix-freebsd-7.dd.gz
466 amd64-windows-server-2003-64bits.dd.gz
467 amd64-windows-server-2008-64bits.dd.gz
468 i386_kvm-linux-debian-lenny
469 i386_kvm-linux-debian-testing
470 i386_kvm-linux-fedora-core13
471 i386_kvm-windows-xp-pro-sp3
472 i386-linux-centos-5.dd.gz
473 i386-linux-debian-etch.dd.gz
474 i386-linux-debian-lenny.dd.gz
475 i386-linux-debian-testing.dd.gz
476 i386-linux-fedora-core10.dd.gz
477 i386-linux-fedora-core11.dd.gz
478 i386-linux-fedora-core12.dd.gz
479 i386-linux-fedora-core13.dd.gz
480 i386-linux-fedora-core7.dd.gz
481 i386-linux-fedora-core8.dd.gz
482 i386-linux-fedora-core9.dd.gz
483 i386-linux-mandriva-2007_springs_powerpack.dd.gz
484 i386-linux-mandriva-2009_powerpack.dd.gz
485 i386-linux-opensuse-11.dd.gz
486 i386-linux-redhatEL-5.0.dd.gz
487 i386-linux-suse-LES10.dd.gz
488 i386-linux-ubuntu-feisty.dd.gz
489 i386-linux-ubuntu-hardy.dd.gz
490 i386-linux-ubuntu-intrepid.dd.gz
491 i386-linux-ubuntu-jaunty.dd.gz
492 i386-linux-ubuntu-karmic.dd.gz
493 i386-linux-ubuntu-lucid.dd.gz
494 i386_mac-mac-osx-server-leopard.dd.gz
495 i386-unix-freebsd-7.dd.gz
496 i386-unix-opensolaris-10.dd.gz
497 i386-unix-opensolaris-11.dd.gz
498 i386-unix-solaris-10.dd.gz
499 ia64-linux-debian-lenny.dd
500 ia64-linux-fedora-core9.dd
501 ia64-linux-redhatEL-5.0.dd
502 x86_64_mac-mac-osx-server-snow-leopard.dd.gz
503 x86_mac-mac-osx-server-snow-leopard.dd.gz
506 Two kind of uses are possible :
508 This command copy your source and execute a configure then a build and finish with tests.
509 bob@caraja:~/Developments/simgrid/tmp_build$ make <name_of_image>
511 This command copy your source and execute a \"ctest -D Experimental\" and submit the result to cdash.
512 bob@caraja:~/Developments/simgrid/tmp_build$ make <name_of_image>_experimental
514 All commands are resumed with :
516 bob@caraja:~/Developments/simgrid/tmp_build$ make pipol_experimental_list_images
517 bob@caraja:~/Developments/simgrid/tmp_build$ make pipol_test_list_images
520 \section faq_installation Installing the SimGrid library with Autotools (valid until V3.3.4)
522 Many people have been asking me questions on how to use SimGrid. Quite
523 often, the questions were not really about SimGrid but on the
524 installation process. This section is intended to help people that are
525 not familiar with compiling C files under UNIX. If you follow these
526 instructions and still have some troubles, drop an e-mail to
527 <simgrid-user@lists.gforge.inria.fr>.
529 \subsection faq_compiling Compiling SimGrid from a stable archive
531 First of all, you need to download the latest version of SimGrid from
532 <a href="http://gforge.inria.fr/frs/?group_id=12">here</a>.
533 Suppose you have uncompressed SimGrid in some temporary location of
534 your home directory (say <tt>/home/joe/tmp/simgrid-3.0.1 </tt>). The
535 simplest way to use SimGrid is to install it in your home
536 directory. Change your directory to
537 <tt>/home/joe/tmp/simgrid-3.0.1</tt> and type
540 ./configure --prefix=$HOME
545 If at some point, something fails, check the section \ref faq_trouble_compil .
546 If it does not help, you can report this problem to the
547 list but, please, avoid sending a laconic mail like "There is a problem. Is it
548 okay?". Send the config.log file which is automatically generated by
549 configure. Try to capture both the standard output and the error output of the
550 <tt>make</tt> command with <tt>script</tt>. There is no way for us to help you
551 without the relevant bits of information.
553 Now, the following directory should have been created :
555 \li <tt>/home/joe/doc/simgrid/html/</tt>
556 \li <tt>/home/joe/lib/</tt>
557 \li <tt>/home/joe/include/</tt>
559 SimGrid is not a binary, it is a library. Both a static and a dynamic
560 version are available. Here is what you can find if you try a <tt>ls
563 \verbatim libsimgrid.a libsimgrid.la libsimgrid.so libsimgrid.so.0 libsimgrid.so.0.0.1
566 Thus, there is two ways to link your program with SimGrid:
567 \li Either you use the static version, e.g
568 \verbatim gcc libsimgrid.a -o MainProgram MainProgram.c
570 In this case, all the SimGrid functions are directly
571 included in <tt>MainProgram</tt> (hence a bigger binary).
572 \li Either you use the dynamic version (the preferred method)
573 \verbatim gcc -lsimgrid -o MainProgram MainProgram.c
575 In this case, the SimGrid functions are not included in
576 <tt>MainProgram</tt> and you need to set your environment
577 variable in such a way that <tt>libsimgrid.so</tt> will be
578 found at runtime. This can be done by adding the following
579 line in your .bashrc (if you use bash and if you have
580 installed the SimGrid libraries in your home directory):
581 \verbatim export LD_LIBRARY_PATH=$HOME/lib/:$LD_LIBRARY_PATH
584 \subsection faq_compiling_java Java bindings don't get compiled
586 The configure script detects automatically whether you have the
587 softwares needed to use the Java bindings or not. At the end of the
588 configure, you can see the configuration picked by the script, which
589 should look similar to
590 \verbatim Configuration of package simgrid' (version 3.3.4-svn) on
593 Compiler: gcc (version: )
595 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
599 Context backend: ucontext
606 In this example, Java backends won't be compiled.
608 On Debian-like systems (which includes ubuntu), you need the following
609 packages: sun-java6-jdk libgcj10-dev. If you cannot find the
610 libgcj10-dev, try another version, like libgcj9-dev (on Ubuntu before
611 9.10) or libgcj11-dev (not released yet, but certainly one day).
612 Please note that you need to activate the contrib and non-free
613 repositories in Debian, and the universe ones in Ubuntu. Java comes at
616 \subsection faq_compiling_snapshoot SimGrid development snapshots
618 We have very high standards on software quality, and we are reluctant releasing
619 a stable release as long as there is still some known bug in the code base. In
620 addition, we added quite an extensive test base, making sure that we correctly
621 test the most important parts of the tool.
623 As an unfortunate conclusion, there may be some time between the stable
624 releases. If you want to benefit from the most recent features we introduced,
625 but don't want to take the risk of an untested version from the SVN, then
626 development snapshots are done for you.
628 These are pre-releases of SimGrid that still fail some tests about features
629 that almost nobody use, or on platforms not being in our core target (which is
630 Linux, Mac, other Unixes and Windows, from the most important to the less
631 one). That means that using this development releases should be safe for most
634 These archives can be found on
635 <a href="http://www.loria.fr/~quinson/Research/SimGrid/">this web page</a>. Once you
636 got the lastest archive, you can compile it just like any archive (see above).
638 \subsection faq_compiling_svn Compiling SimGrid from the SVN
640 The project development takes place in the SVN, where all changes are
641 committed when they happen. Then every once in a while, we make sure that the
642 code quality meets our standard and release an archive from the code in the
643 SVN. We afterward go back to the development in the SVN. So, if you need a
644 recently added feature and can afford some little problem with the stability
645 of the lastest features, you may want to use the SVN version instead of a
648 For that, you first need to get the "simgrid" module from
649 <a href="http://gforge.inria.fr/scm/?group_id=12">here</a>.
651 You won't find any <tt>configure</tt> and a few other things
652 (<tt>Makefile.in</tt>'s, documentation, ...) will be missing as well. The
653 reason for that is that all these files have to be regenerated using the
654 latest versions of <tt>autoconf</tt>, <tt>libtool</tt>, <tt>automake</tt>
655 (>1.9) and <tt>doxygen</tt> (>1.4). To generate the <tt>configure</tt> and
656 the <tt>Makefile.in</tt>'s, you just have to launch the <tt>bootstrap</tt>
657 command that resides in the top of the source tree. Then just follow the
658 instructions of Section \ref faq_compiling.
660 We insist on the fact that you really need the latest versions of
661 autoconf, automake and libtool. Doing this step on exotic architectures/systems
662 (i.e. anything different from a recent linux distribution) may be
663 ... uncertain. If you need to compile the SVN version on a machine where all these
664 dependencies are not met, the easiest is to do <tt>make dist</tt> in the SVN
665 directory of another machine where all dependencies are met. It will create an
666 archive you may deploy on other sites just as a regular stable release.
668 In summary, the following commands will checkout the SVN, regenerate the
669 configure script and friends, configure SimGrid and build it.
671 \verbatim svn checkout svn://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk simgrid
674 ./configure --enable-maintainer-mode --prefix=<where to install SimGrid>
677 Then, if you want to install SimGrid on the current box, just do:
678 \verbatim make install \endverbatim
680 If you want to build an snapshot of the SVN to deploy it on another box (for
681 example because the other machine don't have the autotools), do:
682 \verbatim make dist \endverbatim
684 Moreover, you should never call the autotools manually since you must run
685 them in a specific order with specific arguments. Most of the times, the
686 makefiles will automatically call the tools for you. When it's not possible
687 (such as the first time you checkout the SVN), use the ./bootstrap command
688 to call them explicitly.
691 \subsection faq_setting_MSG Setting up your own MSG code
693 Do not build your simulator by modifying the SimGrid examples. Go
694 outside the SimGrid source tree and create your own working directory
695 (say <tt>/home/joe/SimGrid/MyFirstScheduler/</tt>).
697 Suppose your simulation has the following structure (remember it is
698 just an example to illustrate a possible way to compile everything;
699 feel free to organize it as you want).
701 \li <tt>sched.h</tt>: a description of the core of the
702 scheduler (i.e. which functions are can be used by the
703 agents). For example we could find the following functions
704 (master, forwarder, slave).
706 \li <tt>sched.c</tt>: a C file including <tt>sched.h</tt> and
707 implementing the core of the scheduler. Most of these
708 functions use the MSG functions defined in section \ref
711 \li <tt>masterslave.c</tt>: a C file with the main function, i.e.
712 the MSG initialization (MSG_global_init()), the platform
713 creation (e.g. with MSG_create_environment()), the
714 deployment phase (e.g. with MSG_function_register() and
715 MSG_launch_application()) and the call to
718 To compile such a program, we suggest to use the following
719 Makefile. It is a generic Makefile that we have used many times with
720 our students when we teach the C language.
724 masterslave: masterslave.o sched.o
726 INSTALL_PATH = $$HOME
728 PEDANTIC_PARANOID_FREAK = -O0 -Wshadow -Wcast-align \
729 -Waggregate-return -Wmissing-prototypes -Wmissing-declarations \
730 -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations \
731 -Wmissing-noreturn -Wredundant-decls -Wnested-externs \
732 -Wpointer-arith -Wwrite-strings -finline-functions
733 REASONABLY_CAREFUL_DUDE = -Wall
734 NO_PRAYER_FOR_THE_WICKED = -w -O2
735 WARNINGS = $(REASONABLY_CAREFUL_DUDE)
736 CFLAGS = -g $(WARNINGS)
738 INCLUDES = -I$(INSTALL_PATH)/include
739 DEFS = -L$(INSTALL_PATH)/lib/
740 LDADD = -lm -lsimgrid
744 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) $^ $(LIBS) $(LDADD) -o $@
747 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) -c -o $@ $<
750 rm -f $(BIN_FILES) *.o *~
756 The first two lines indicates what should be build when typing make
757 (<tt>masterslave</tt>) and of which files it is to be made of
758 (<tt>masterslave.o</tt> and <tt>sched.o</tt>). This makefile assumes
759 that you have set up correctly your <tt>LD_LIBRARY_PATH</tt> variable
760 (look, there is a <tt>LDADD = -lm -lsimgrid</tt>). If you prefer using
761 the static version, remove the <tt>-lsimgrid</tt> and add a
762 <tt>$(INSTALL_PATH)/lib/libsimgrid.a</tt> on the next line, right
763 after the <tt>LIBS = </tt>.
765 More generally, if you have never written a Makefile by yourself, type
766 in a terminal : <tt>info make</tt> and read the introduction. The
767 previous example should be enough for a first try but you may want to
768 perform some more complex compilations...
770 \subsection faq_setting_GRAS Setting up your own GRAS code
772 If you use the GRAS interface instead of the MSG one, then previous section
773 is not the better source of information. Instead, you should check the GRAS
774 tutorial in general, and the \ref GRAS_tut_tour_setup in particular.
776 \section faq_howto Feature related questions
778 \subsection faq_MIA "Could you please add (your favorite feature here) to SimGrid?"
780 Here is the deal. The whole SimGrid project (MSG, SURF, GRAS, ...) is
781 meant to be kept as simple and generic as possible. We cannot add
782 functions for everybody's needs when these functions can easily be
783 built from the ones already in the API. Most of the time, it is
784 possible and when it was not possible we always have upgraded the API
785 accordingly. When somebody asks us a question like "How to do that?
786 Is there a function in the API to simply do this?", we're always glad
787 to answer and help. However if we don't need this code for our own
788 need, there is no chance we're going to write it... it's your job! :)
789 The counterpart to our answers is that once you come up with a neat
790 implementation of this feature (task duplication, RPC, thread
791 synchronization, ...), you should send it to us and we will be glad to
792 add it to the distribution. Thus, other people will take advantage of
793 it (and we don't have to answer this question again and again ;).
795 You'll find in this section a few "Missing In Action" features. Many
796 people have asked about it and we have given hints on how to simply do
797 it with MSG. Feel free to contribute...
799 \subsection faq_MIA_MSG MSG features
801 \subsubsection faq_MIA_examples I want some more complex MSG examples!
803 Many people have come to ask me a more complex example and each time,
804 they have realized afterward that the basics were in the previous three
807 Of course they have often been needing more complex functions like
808 MSG_process_suspend(), MSG_process_resume() and
809 MSG_process_isSuspended() (to perform synchronization), or
810 MSG_task_Iprobe() and MSG_process_sleep() (to avoid blocking
811 receptions), or even MSG_process_create() (to design asynchronous
812 communications or computations). But the examples are sufficient to
815 We know. We should add some more examples, but not really some more
816 complex ones... We should add some examples that illustrate some other
817 functionalists (like how to simply encode asynchronous
818 communications, RPC, process migrations, thread synchronization, ...)
819 and we will do it when we will have a little bit more time. We have
820 tried to document the examples so that they are understandable. Tell
821 us if something is not clear and once again feel free to participate!
824 \subsubsection faq_MIA_taskdup Missing in action: MSG Task duplication/replication
826 There is no task duplication in MSG. When you create a task, you can
827 process it or send it somewhere else. As soon as a process has sent
828 this task, he doesn't have this task anymore. It's gone. The receiver
829 process has got the task. However, you could decide upon receiving to
830 create a "copy" of a task but you have to handle by yourself the
831 semantic associated to this "duplication".
833 As we already told, we prefer keeping the API as simple as
834 possible. This kind of feature is rather easy to implement by users
835 and the semantic you associate really depends on people. Having a
836 *generic* task duplication mechanism is not that trivial (in
837 particular because of the data field). That is why I would recommand
838 that you write it by yourself even if I can give you advice on how to
841 You have the following functions to get informations about a task:
842 MSG_task_get_name(), MSG_task_get_compute_duration(),
843 MSG_task_get_remaining_computation(), MSG_task_get_data_size(),
844 and MSG_task_get_data().
846 You could use a dictionary (#xbt_dict_t) of dynars (#xbt_dynar_t). If
847 you still don't see how to do it, please come back to us...
849 \subsubsection faq_MIA_asynchronous I want to do asynchronous communications in MSG
851 In the past (version <= 3.4), there was no function to perform asynchronous communications.
852 It could easily be implemented by creating new process when needed though. Since version 3.5,
853 we have introduced the following functions:
862 We refer you to the description of these functions for more details on their usage as well
863 as to the exemple section on \ref MSG_ex_asynchronous_communications.
865 \subsubsection faq_MIA_thread_synchronization I need to synchronize my MSG processes
867 You obviously cannot use pthread_mutexes of pthread_conds since we handle every
868 scheduling related decision within SimGrid.
870 In the past (version <=3.3.4) you could do it by playing with
871 MSG_process_suspend() and MSG_process_resume() or with fake communications (using MSG_task_get(),
872 MSG_task_put() and MSG_task_Iprobe()).
874 Since version 3.4, you can use classical synchronization structures. See page \ref XBT_synchro or simply check in
875 include/xbt/synchro_core.h.
877 \subsubsection faq_MIA_host_load Where is the get_host_load function hidden in MSG?
879 There is no such thing because its semantic wouldn't be really
880 clear. Of course, it is something about the amount of host throughput,
881 but there is as many definition of "host load" as people asking for
882 this function. First, you have to remember that resource availability
883 may vary over time, which make any load notion harder to define.
885 It may be instantaneous value or an average one. Moreover it may be only the
886 power of the computer, or may take the background load into account, or may
887 even take the currently running tasks into account. In some SURF models,
888 communications have an influence on computational power. Should it be taken
891 First of all, it's near to impossible to predict the load beforehands in the
892 simulator since it depends on too much parameters (background load
893 variation, bandwidth sharing algorithmic complexity) some of them even being
894 not known beforehands (other task starting at the same time). So, getting
895 this information is really hard (just like in real life). It's not just that
896 we want MSG to be as painful as real life. But as it is in some way
897 realistic, we face some of the same problems as we would face in real life.
899 How would you do it for real? The most common option is to use something
900 like NWS that performs active probes. The best solution is probably to do
901 the same within MSG, as in next code snippet. It is very close from what you
902 would have to do out of the simulator, and thus gives you information that
903 you could also get in real settings to not hinder the realism of your
907 double get_host_load() {
908 m_task_t task = MSG_task_create("test", 0.001, 0, NULL);
909 double date = MSG_get_clock();
911 MSG_task_execute(task);
912 date = MSG_get_clock() - date;
913 MSG_task_destroy(task);
918 Of course, it may not match your personal definition of "host load". In this
919 case, please detail what you mean on the mailing list, and we will extend
920 this FAQ section to fit your taste if possible.
922 \subsubsection faq_MIA_communication_time How can I get the *real* communication time?
924 Communications are synchronous and thus if you simply get the time
925 before and after a communication, you'll only get the transmission
926 time and the time spent to really communicate (it will also take into
927 account the time spent waiting for the other party to be
928 ready). However, getting the *real* communication time is not really
929 hard either. The following solution is a good starting point.
934 m_task_t task = MSG_task_create("Task", task_comp_size, task_comm_size,
935 calloc(1,sizeof(double)));
936 *((double*) task->data) = MSG_get_clock();
937 MSG_task_put(task, slaves[i % slaves_count], PORT_22);
938 XBT_INFO("Send completed");
943 m_task_t task = NULL;
946 time1 = MSG_get_clock();
947 a = MSG_task_get(&(task), PORT_22);
948 time2 = MSG_get_clock();
949 if(time1<*((double *)task->data))
950 time1 = *((double *) task->data);
951 XBT_INFO("Communication time : \"%f\" ", time2-time1);
953 MSG_task_destroy(task);
958 \subsection faq_MIA_SimDag SimDag related questions
960 \subsubsection faq_SG_comm Implementing communication delays between tasks.
962 A classic question of SimDag newcomers is about how to express a
963 communication delay between tasks. The thing is that in SimDag, both
964 computation and communication are seen as tasks. So, if you want to
965 model a data dependency between two DAG tasks t1 and t2, you have to
966 create 3 SD_tasks: t1, t2 and c and add dependencies in the following
970 SD_task_dependency_add(NULL, NULL, t1, c);
971 SD_task_dependency_add(NULL, NULL, c, t2);
974 This way task t2 cannot start before the termination of communication c
975 which in turn cannot start before t1 ends.
977 When creating task c, you have to associate an amount of data (in bytes)
978 corresponding to what has to be sent by t1 to t2.
980 Finally to schedule the communication task c, you have to build a list
981 comprising the workstations on which t1 and t2 are scheduled (w1 and w2
982 for example) and build a communication matrix that should look like
985 \subsubsection faq_SG_DAG How to implement a distributed dynamic scheduler of DAGs.
987 Distributed is somehow "contagious". If you start making distributed
988 decisions, there is no way to handle DAGs directly anymore (unless I
989 am missing something). You have to encode your DAGs in term of
990 communicating process to make the whole scheduling process
991 distributed. Here is an example of how you could do that. Assume T1
992 has to be done before T2.
995 int your_agent(int argc, char *argv[] {
997 T1 = MSG_task_create(...);
998 T2 = MSG_task_create(...);
1002 if(cond) MSG_task_execute(T1);
1004 if((MSG_task_get_remaining_computation(T1)=0.0) && (you_re_in_a_good_mood))
1005 MSG_task_execute(T2)
1007 /* do something else */
1013 If you decide that the distributed part is not that much important and that
1014 DAG is really the level of abstraction you want to work with, then you should
1015 give a try to \ref SD_API.
1017 \subsection faq_MIA_generic Generic features
1019 \subsubsection faq_more_processes Increasing the amount of simulated processes
1021 Here are a few tricks you can apply if you want to increase the amount
1022 of processes in your simulations.
1024 - <b>A few thousands of simulated processes</b> (soft tricks)\n
1025 SimGrid can use either pthreads library or the UNIX98 contextes. On
1026 most systems, the number of pthreads is limited and then your
1027 simulation may be limited for a stupid reason. This is especially
1028 true with the current linux pthreads, and I cannot get more than
1029 2000 simulated processes with pthreads on my box. The UNIX98
1030 contexts allow me to raise the limit to 25,000 simulated processes
1032 The <tt>--with-context</tt> option of the <tt>./configure</tt>
1033 script allows you to choose between UNIX98 contextes
1034 (<tt>--with-context=ucontext</tt>) and the pthread version
1035 (<tt>--with-context=pthread</tt>). The default value is ucontext
1036 when the script detect a working UNIX98 context implementation. On
1037 Windows boxes, the provided value is discarded and an adapted
1038 version is picked up.\n\n
1039 We experienced some issues with contextes on some rare systems
1040 (solaris 8 and lower or old alpha linuxes comes to mind). The main
1041 problem is that the configure script detect the contextes as being
1042 functional when it's not true. If you happen to use such a system,
1043 switch manually to the pthread version, and provide us with a good
1044 patch for the configure script so that it is done automatically ;)
1046 - <b>Hundred thousands of simulated processes</b> (hard-core tricks)\n
1047 As explained above, SimGrid can use UNIX98 contextes to represent
1048 and handle the simulated processes. Thanks to this, the main
1049 limitation to the number of simulated processes becomes the
1050 available memory.\n\n
1051 Here are some tricks I had to use in order to run a token ring
1052 between 25,000 processes on my laptop (1Gb memory, 1.5Gb swap).\n
1053 - First of all, make sure your code runs for a few hundreds
1054 processes before trying to push the limit. Make sure it's
1055 valgrind-clean, ie that valgrind does not report neither memory
1056 error nor memory leaks. Indeed, numerous simulated processes
1057 result in *fat* simulation hindering debugging.
1058 - It was really boring to write 25,000 entries in the deployment
1059 file, so I wrote a little script
1060 <tt>examples/gras/mutual_exclusion/simple_token/make_deployment.pl</tt>, which you may
1061 want to adapt to your case. You could also think about hijacking
1062 the SURFXML parser (have look at \ref faq_flexml_bypassing).
1063 - The deployment file became quite big, so I had to do what is in
1064 the FAQ entry \ref faq_flexml_limit
1065 - Each UNIX98 context has its own stack entry. As debugging this is
1066 quite hairly, the default value is a bit overestimated so that
1067 user don't get into trouble about this. You want to tune this
1068 size to increse the number of processes. This is the
1069 <tt>STACK_SIZE</tt> define in
1070 <tt>src/xbt/xbt_context_sysv.c</tt>, which is 128kb by default.
1071 Reduce this as much as you can, but be warned that if this value
1072 is too low, you'll get a segfault. The token ring example, which
1073 is quite simple, runs with 40kb stacks.
1074 - You may tweak the logs to reduce the stack size further. When
1075 logging something, we try to build the string to display in a
1076 char array on the stack. The size of this array is constant (and
1077 equal to XBT_LOG_BUFF_SIZE, defined in include/xbt/log/h). If the
1078 string is too large to fit this buffer, we move to a dynamically
1079 sized buffer. In which case, we have to traverse one time the log
1080 event arguments to compute the size we need for the buffer,
1081 malloc it, and traverse the argument list again to do the actual
1083 The idea here is to move XBT_LOG_BUFF_SIZE to 1, forcing the logs
1084 to use a dynamic array each time. This allows us to lower further
1085 the stack size at the price of some performance loss...\n
1086 This allowed me to run the reduce the stack size to ... 4k. Ie,
1087 on my 1Gb laptop, I can run more than 250,000 processes!
1089 \subsubsection faq_MIA_batch_scheduler Is there a native support for batch schedulers in SimGrid?
1091 No, there is no native support for batch schedulers and none is
1092 planned because this is a very specific need (and doing it in a
1093 generic way is thus very hard). However some people have implemented
1094 their own batch schedulers. Vincent Garonne wrote one during his PhD
1095 and put his code in the contrib directory of our SVN so that other can
1096 keep working on it. You may find inspiring ideas in it.
1098 \subsubsection faq_MIA_checkpointing I need a checkpointing thing
1100 Actually, it depends on whether you want to checkpoint the simulation, or to
1101 simulate checkpoints.
1103 The first one could help if your simulation is a long standing process you
1104 want to keep running even on hardware issues. It could also help to
1105 <i>rewind</i> the simulation by jumping sometimes on an old checkpoint to
1106 cancel recent calculations.\n
1107 Unfortunately, such thing will probably never exist in SG. One would have to
1108 duplicate all data structures because doing a rewind at the simulator level
1109 is very very hard (not talking about the malloc free operations that might
1110 have been done in between). Instead, you may be interested in the Libckpt
1111 library (http://www.cs.utk.edu/~plank/plank/www/libckpt.html). This is the
1112 checkpointing solution used in the condor project, for example. It makes it
1113 easy to create checkpoints (at the OS level, creating something like core
1114 files), and rerunning them on need.
1116 If you want to simulate checkpoints instead, it means that you want the
1117 state of an executing task (in particular, the progress made towards
1118 completion) to be saved somewhere. So if a host (and the task executing on
1119 it) fails (cf. #MSG_HOST_FAILURE), then the task can be restarted
1120 from the last checkpoint.\n
1122 Actually, such a thing does not exists in SimGrid either, but it's just
1123 because we don't think it is fundamental and it may be done in the user code
1124 at relatively low cost. You could for example use a watcher that
1125 periodically get the remaining amount of things to do (using
1126 MSG_task_get_remaining_computation()), or fragment the task in smaller
1129 \subsection faq_platform Platform building and Dynamic resources
1131 \subsubsection faq_platform_example Where can I find SimGrid platform files?
1133 There is several little examples in the archive, in the examples/msg
1134 directory. From time to time, we are asked for other files, but we
1135 don't have much at hand right now.
1137 You should refer to the Platform Description Archive
1138 (http://pda.gforge.inria.fr) project to see the other platform file we
1139 have available, as well as the Simulacrum simulator, meant to generate
1140 SimGrid platforms using all classical generation algorithms.
1142 \subsubsection faq_platform_alnem How can I automatically map an existing platform?
1144 We are working on a project called ALNeM (Application-Level Network
1145 Mapper) which goal is to automatically discover the topology of an
1146 existing network. Its output will be a platform description file
1147 following the SimGrid syntax, so everybody will get the ability to map
1148 their own lab network (and contribute them to the catalog project).
1149 This tool is not ready yet, but it move quite fast forward. Just stay
1152 \subsubsection faq_platform_synthetic Generating synthetic but realistic platforms
1154 The third possibility to get a platform file (after manual or
1155 automatic mapping of real platforms) is to generate synthetic
1156 platforms. Getting a realistic result is not a trivial task, and
1157 moreover, nobody is really able to define what "realistic" means when
1158 speaking of topology files. You can find some more thoughts on this
1160 <a href="http://graal.ens-lyon.fr/~alegrand/articles/Simgrid-Introduction.pdf">slides</a>.
1162 If you are looking for an actual tool, there we have a little tool to
1163 annotate Tiers-generated topologies. This perl-script is in
1164 <tt>tools/platform_generation/</tt> directory of the SVN. Dinda et Al.
1165 released a very comparable tool, and called it GridG.
1167 \subsubsection faq_SURF_multicore Modeling multi-core resources
1169 There is currently no native support for multi-core or SMP machines in
1170 SimGrid. We are currently working on it, but coming up with the right
1171 model is very hard: Cores share caches and bus to access memory and
1172 thus interfere with each others. Memory contention is a crucial
1173 component of multi-core modeling.
1175 In the meanwhile, some user-level tricks can reveal sufficient for
1176 you. For example, you may model each core by a CPU and add some very
1177 high speed links between them. This complicates a bit the user code
1178 since you have to remember that when you assign something to a (real)
1179 host, it can be any of the (fake) hosts representing the cores of a
1180 given machine. For that, you can use the prop tag of the XML files as
1181 follows. Your code should then look at the ‘machine’ property
1182 associated with each workstation, and run parallel tasks over all
1183 cores of the machine.
1186 <host id="machine0/core0" power="91500E6">
1187 <prop id="machine" value="machine0"/>
1188 <prop id="core" value="0"/>
1190 <host id="machine0/core1" power="91500E6">
1191 <prop id="machine" value="machine0"/>
1192 <prop id="core" value="1"/>
1198 \subsubsection faq_SURF_dynamic Modeling dynamic resource availability
1200 A nice feature of SimGrid is that it enables you to seamlessly have
1201 resources whose availability change over time. When you build a
1202 platform, you generally declare hosts like that:
1205 <host id="host A" power="100.00"/>
1208 If you want the availability of "host A" to change over time, the only
1209 thing you have to do is change this definition like that:
1212 <host id="host A" power="100.00" availability_file="trace_A.txt" state_file="trace_A_failure.txt"/>
1215 For hosts, availability files are expressed in fraction of available
1216 power. Let's have a look at what "trace_A.txt" may look like:
1225 At time 0, our host will deliver 100 flop/s. At time 11.0, it will
1226 deliver only 50 flop/s until time 20.0 where it will will start
1227 delivering 90 flop/s. Last at time 21.0 (20.0 plus the periodicity
1228 1.0), we'll be back to the beginning and it will deliver 100 flop/s.
1230 Now let's look at the state file:
1237 A negative value means "off" while a positive one means "on". At time
1238 1.0, the host is on. At time 1.0, it is turned off and at time 2.0, it
1239 is turned on again until time 12 (2.0 plus the periodicity 10.0). It
1240 will be turned on again at time 13.0 until time 23.0, and so on.
1242 Now, let's look how the same kind of thing can be done for network
1243 links. A usual declaration looks like:
1246 <link id="LinkA" bandwidth="10.0" latency="0.2"/>
1249 You have at your disposal the following options: bandwidth_file,
1250 latency_file and state_file. The only difference with hosts is that
1251 bandwidth_file and latency_file do not express fraction of available
1252 power but are expressed directly in bytes per seconds and seconds.
1254 \subsubsection faq_platform_multipath How to express multipath routing in platform files?
1256 It is unfortunately impossible to express the fact that there is more
1257 than one routing path between two given hosts. Let's consider the
1258 following platform file:
1261 <route src="A" dst="B">
1264 <route src="B" dst="C">
1267 <route src="A" dst="C">
1272 Although it is perfectly valid, it does not mean that data traveling
1273 from A to C can either go directly (using link 3) or through B (using
1274 links 1 and 2). It simply means that the routing on the graph is not
1275 trivial, and that data do not following the shortest path in number of
1276 hops on this graph. Another way to say it is that there is no implicit
1277 in these routing descriptions. The system will only use the routes you
1278 declare (such as <route src="A" dst="C"><link:ctn
1279 id="3"/></route>), without trying to build new routes by aggregating
1282 You are also free to declare platform where the routing is not
1283 symmetric. For example, add the following to the previous file:
1286 <route src="C" dst="A">
1292 This makes sure that data from C to A go through B where data from A
1293 to C go directly. Don't worry about realism of such settings since
1294 we've seen ways more weird situation in real settings (in fact, that's
1295 the realism of very regular platforms which is questionable, but
1296 that's another story).
1298 \subsubsection faq_flexml_bypassing Bypassing the XML parser with your own C functions
1300 So you want to bypass the XML files parser, uh? Maybe doing some parameter
1301 sweep experiments on your simulations or so? This is possible, and
1302 it's not even really difficult (well. Such a brutal idea could be
1303 harder to implement). Here is how it goes.
1305 For this, you have to first remember that the XML parsing in SimGrid is done
1306 using a tool called FleXML. Given a DTD, this gives a flex-based parser. If
1307 you want to bypass the parser, you need to provide some code mimicking what
1308 it does and replacing it in its interactions with the SURF code. So, let's
1309 have a look at these interactions.
1311 FleXML parser are close to classical SAX parsers. It means that a
1312 well-formed SimGrid platform XML file might result in the following
1315 - start "platform_description" with attribute version="2"
1316 - start "host" with attributes id="host1" power="1.0"
1318 - start "host" with attributes id="host2" power="2.0"
1320 - start "link" with ...
1322 - start "route" with ...
1323 - start "link:ctn" with ...
1326 - end "platform_description"
1328 The communication from the parser to the SURF code uses two means:
1329 Attributes get copied into some global variables, and a surf-provided
1330 function gets called by the parser for each event. For example, the event
1331 - start "host" with attributes id="host1" power="1.0"
1333 let the parser do something roughly equivalent to:
1335 strcpy(A_host_id,"host1");
1340 In SURF, we attach callbacks to the different events by initializing the
1341 pointer functions to some the right surf functions. Since there can be
1342 more than one callback attached to the same event (if more than one
1343 model is in use, for example), they are stored in a dynar. Example in
1344 workstation_ptask_L07.c:
1346 /* Adding callback functions */
1347 surf_parse_reset_parser();
1348 surfxml_add_callback(STag_surfxml_host_cb_list, &parse_cpu_init);
1349 surfxml_add_callback(STag_surfxml_prop_cb_list, &parse_properties);
1350 surfxml_add_callback(STag_surfxml_link_cb_list, &parse_link_init);
1351 surfxml_add_callback(STag_surfxml_route_cb_list, &parse_route_set_endpoints);
1352 surfxml_add_callback(ETag_surfxml_link_c_ctn_cb_list, &parse_route_elem);
1353 surfxml_add_callback(ETag_surfxml_route_cb_list, &parse_route_set_route);
1355 /* Parse the file */
1356 surf_parse_open(file);
1357 xbt_assert(!surf_parse(), "Parse error in %s", file);
1361 So, to bypass the FleXML parser, you need to write your own version of the
1362 surf_parse function, which should do the following:
1363 - Fill the A_<tag>_<attribute> variables with the wanted values
1364 - Call the corresponding STag_<tag>_fun function to simulate tag start
1365 - Call the corresponding ETag_<tag>_fun function to simulate tag end
1366 - (do the same for the next set of values, and loop)
1368 Then, tell SimGrid that you want to use your own "parser" instead of the stock one:
1370 surf_parse = surf_parse_bypass_environment;
1371 MSG_create_environment(NULL);
1372 surf_parse = surf_parse_bypass_application;
1373 MSG_launch_application(NULL);
1376 A set of macros are provided at the end of
1377 include/surf/surfxml_parse.h to ease the writing of the bypass
1378 functions. An example of this trick is distributed in the file
1379 examples/msg/masterslave/masterslave_bypass.c
1381 \subsection faq_simgrid_configuration Changing SimGrid's behavior
1383 A number of options can be given at runtime to change the default
1384 SimGrid behavior. In particular, you can change the default cpu and
1387 \subsubsection faq_simgrid_configuration_fullduplex Using Fullduplex
1389 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
1390 of interconnected hosts, see an example here:
1392 simgrid_svn_sources/exemples/msg/gtnets/fullduplex-p.xml
1395 Using fullduplex support ongoing and incoming communication flows are
1396 treated independently for most models. The exception is the LV08 model which
1397 adds 0.05 of usage on the opposite direction for each new created flow. This
1398 can be useful to simulate some important TCP phenomena such as ack compression.
1400 Running a fullduplex example:
1402 cd simgrid_svn_sources/exemples/msg/gtnets
1403 ./gtnets fullduplex-p.xml fullduplex-d.xml --cfg=fullduplex:1
1410 \subsubsection faq_simgrid_configuration_gtnets Using GTNetS
1412 It is possible to use a packet-level network simulator
1413 instead of the default flow-based simulation. You may want to use such
1414 an approach if you have doubts about the validity of the default model
1415 or if you want to perform some validation experiments. At the moment,
1416 we support the GTNetS simulator (it is still rather experimental
1417 though, so leave us a message if you play with it).
1421 To enable GTNetS model inside SimGrid it is needed to patch the GTNetS simulator source code
1422 and build/install it from scratch
1425 - <b>Download and enter the recent downloaded GTNetS directory</b>
1428 svn checkout svn://scm.gforge.inria.fr/svn/simgrid/contrib/trunk/GTNetS/
1433 - <b>Use the following commands to unzip and patch GTNetS package to work within SimGrid.</b>
1436 unzip gtnets-current.zip
1437 tar zxvf gtnets-current-patch.tgz
1439 cat ../00*.patch | patch -p1
1442 - <b>OPTIONALLY</b> you can use a patch for itanium 64bit processor family.
1445 cat ../AMD64-FATAL-Removed-DUL_SIZE_DIFF-Added-fPIC-compillin.patch | patch -p1
1448 - <b>Compile GTNetS</b>
1450 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.
1454 ln -sf Makefile.linux Makefile
1460 - <b>NOTE</b> A lot of warnings are expected but the application should compile
1461 just fine. If the makefile insists in compiling some QT libraries
1462 please try a make clean before asking for help.
1465 - <b>To compile optimized version</b>
1472 - <b>Installing GTNetS</b>
1474 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.
1477 ln -sf /<absolute_path>/gtnets_current/libgtsim-debug.so /<userhome>/usr/lib/libgtnets.so
1478 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/<userhome>/usr/lib/libgtnets.so
1479 mkdir /<userhome>/usr/include/gtnets
1480 cp -fr SRC/*.h /<userhome>/usr/include/gtnets
1484 - <b>Enable GTNetS support in SimGrid</b>
1486 In order to enable gtnets with simgrid you have to give where is gtnets. (path to \<gtnets_path\>/lib and \<gtnets_path\>/include)
1489 Since v3.4 (with cmake)
1490 cmake . -Dgtnets_path=/<userhome>/usr
1492 Until v3.4 (with autotools)
1493 ./configure --with-gtnets=/<userhome>/usr
1496 - <b>Once you have followed all the instructions for compiling and
1497 installing successfully you can activate this feature at
1498 runntime with the following options:</b>
1501 Since v3.4 (with cmake)
1506 Until v3.4 (with autotools)
1507 cd simgrid/example/msg/
1513 - <b>Or try the GTNetS model dogbone example with</b>
1516 gtnets/gtnets gtnets/onelink-p.xml gtnets/onelink-d.xml --cfg=network_model:GTNets
1520 A long version of this <a href="http://gforge.inria.fr/docman/view.php/12/6283/GTNetS HowTo.html">HowTo</a> it is available
1523 More about GTNetS simulator at <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/index.html">GTNetS Website</a>
1527 The patches provided by us worked successfully with GTNetS found
1528 <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/software/gtnets-current.zip">here</a>,
1529 dated from 12th June 2008. Due to the discontinuing development of
1530 GTNetS it is impossible to precise a version number. We STRONGLY recommend you
1531 to download and install the GTNetS version found in SimGrid repository as explained above.
1536 \subsubsection faq_simgrid_configuration_alternate_network Using alternative flow models
1538 The default simgrid network model uses a max-min based approach as
1539 explained in the research report
1540 <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>.
1541 Other models have been proposed and implemented since then (see for example
1542 <a href="http://mescal.imag.fr/membres/arnaud.legrand/articles/simutools09.pdf">Accuracy Study and Improvement of Network Simulation in the SimGrid Framework</a>)
1543 and can be activated at runtime. For example:
1545 ./mycode platform.xml deployment.xml --cfg=workstation/model:compound --cfg=network/model:LV08 -cfg=cpu/model:Cas01
1548 Possible models for the network are currently "Constant", "CM02",
1549 "LegrandVelho", "GTNets", Reno", "Reno2", "Vegas". Others will
1550 probably be added in the future and many of the previous ones are
1551 experimental and are likely to disappear without notice... To know the
1552 list of the currently implemented models, you should use the
1553 --help-models command line option.
1556 ./masterslave_forwarder ../small_platform.xml deployment_masterslave.xml --help-models
1557 Long description of the workstation models accepted by this simulator:
1558 CLM03: Default workstation model, using LV08 and CM02 as network and CPU
1559 compound: Workstation model allowing you to use other network and CPU models
1560 ptask_L07: Workstation model with better parallel task modeling
1561 Long description of the CPU models accepted by this simulator:
1562 Cas01_fullupdate: CPU classical model time=size/power
1563 Cas01: Variation of Cas01_fullupdate with partial invalidation optimization of lmm system. Should produce the same values, only faster
1564 CpuTI: Variation of Cas01 with also trace integration. Should produce the same values, only faster if you use availability traces
1565 Long description of the network models accepted by this simulator:
1566 Constant: Simplistic network model where all communication take a constant time (one second)
1567 CM02: Realistic network model with lmm_solve and no correction factors
1568 LV08: Realistic network model with lmm_solve and these correction factors: latency*=10.4, bandwidth*=.92, S=8775
1569 Reno: Model using lagrange_solve instead of lmm_solve (experts only)
1570 Reno2: Model using lagrange_solve instead of lmm_solve (experts only)
1571 Vegas: Model using lagrange_solve instead of lmm_solve (experts only)
1574 \subsection faq_tracing Tracing Simulations for Visualization
1576 The trace visualization is widely used to observe and understand the behavior
1577 of parallel applications and distributed algorithms. Usually, this is done in a
1578 two-step fashion: the user instruments the application and the traces are
1579 analyzed after the end of the execution. The visualization itself can highlights
1580 unexpected behaviors, bottlenecks and sometimes can be used to correct
1581 distributed algorithms. The SimGrid team has instrumented the library
1582 in order to let users trace their simulations and analyze them. This part of the
1583 user manual explains how the tracing-related features can be enabled and used
1584 during the development of simulators using the SimGrid library.
1586 \subsubsection faq_tracing_howitworks How it works
1588 For now, the SimGrid library is instrumented so users can trace the <b>platform
1589 utilization</b> using the MSG, SimDAG and SMPI interface. This means that the tracing will
1590 register how much power is used for each host and how much bandwidth is used for
1591 each link of the platform. The idea with this type of tracing is to observe the
1592 overall view of resources utilization in the first place, especially the
1593 identification of bottlenecks, load-balancing among hosts, and so on.
1595 The idea of the tracing facilities is to give SimGrid users to possibility to
1596 classify MSG and SimDAG tasks by category, tracing the platform utilization
1597 (hosts and links) for each of the categories. For that,
1598 the tracing interface enables the declaration of categories and a function to
1599 mark a task with a previously declared category. <em>The tasks that are not
1600 classified according to a category are not traced</em>. Even if the user
1601 does not specify any category, the simulations can still be traced in terms
1602 of resource utilization by using a special parameter that is detailed below.
1604 \subsubsection faq_tracing_enabling Enabling using CMake
1606 With the sources of SimGrid, it is possible to enable the tracing
1607 using the parameter <b>-Denable_tracing=ON</b> when the cmake is executed.
1608 The section \ref faq_tracing_functions describes all the functions available
1609 when this Cmake options is activated. These functions will have no effect
1610 if SimGrid is configured without this option (they are wiped-out by the
1614 $ cmake -Denable_tracing=ON .
1618 \subsubsection faq_tracing_functions Tracing Functions
1620 \li <b>\c TRACE_category (const char *category)</b>: This function should be used
1621 to define a user category. The category can be used to differentiate the tasks
1622 that are created during the simulation (for example, tasks from server1,
1623 server2, or request tasks, computation tasks, communication tasks).
1624 All resource utilization (host power and link bandwidth) will be
1625 classified according to the task category. Tasks that do not belong to a
1626 category are not traced. The color for the category that is being declared
1627 is random (use next function to specify a color).
1629 \li <b>\c TRACE_category_with_color (const char *category, const char *color)</b>: Same
1630 as TRACE_category, but let user specify a color encoded as a RGB-like string with
1631 three floats from 0 to 1. So, to specify a red color, the user can pass "1 0 0" as
1632 color parameter. A light-gray color can be specified using "0.7 0.7 0.7" as color.
1634 \li <b>\c TRACE_msg_set_task_category (m_task_t task, const char *category)</b>:
1635 This function should be called after the creation of a MSG task, to define the
1636 category of that task. The first parameter \c task must contain a task that was
1637 created with the function \c MSG_task_create. The second parameter
1638 \c category must contain a category that was previously defined by the function
1641 \li <b>\c TRACE_sd_set_task_category (SD_task_t task, const char *category)</b>:
1642 This function should be called after the creation of a SimDAG task, to define the
1643 category of that task. The first parameter \c task must contain a task that was
1644 created with the function \c MSG_task_create. The second parameter
1645 \c category must contain a category that was previously defined by the function
1648 \li <b>\c TRACE_[host|link]_variable_declare (const char *variable)</b>:
1649 Declare a user variable that will be associated to host/link. A variable can
1650 be used to trace user variables such as the number of tasks in a server,
1651 the number of clients in an application (for hosts), and so on.
1653 \li <b>\c TRACE_[host|link]_variable_[set|add|sub] (const char *[host|link], const char *variable, double value)</b>:
1654 Set the value of a given user variable for a given host/link. The value
1655 of this variable is always associated to the host/link. The host/link
1656 parameters should be its name as the one listed in the platform file.
1658 \li <b>\c TRACE_[host|link]_variable_[set|add|sub]_with_time (double time, const char *[host|link], const char *variable, double value)</b>:
1659 Same as TRACE_[host|link]_variable_[set|add|sub], but let user specify
1660 the time used to trace it. Users can specify a time that is not the
1661 simulated clock time as defined by the core simulator. This allows
1662 a fine-grain control of time definition, but should be used with
1663 caution since the trace can be inconsistent if resource utilization
1664 traces are also traced.
1666 \li <b>\c TRACE_link_srcdst_variable_[set|add|sub] (const char *src, const char *dst, const char *variable, double value)</b>:
1667 Same as TRACE_link_variable_[set|add|sub], but now users specify a source and
1668 destination hosts (as the names from the platform file). The tracing library
1669 will get the corresponding route that connects those two hosts (src and dst) and
1670 [set|add|sub] the value's variable for all the links of the route.
1672 \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>:
1673 Same as TRACE_link_srcdst_variable_[set|add|sub], but user specify a time different from the simulated time.
1675 \subsubsection faq_tracing_options Tracing configuration Options
1677 These are the options accepted by the tracing system of SimGrid:
1682 Safe switch. It activates (or deactivates) the tracing system.
1683 No other tracing options take effect if this one is not activated.
1688 Register the simulation platform in the trace file.
1691 tracing/onelink_only
1693 By default, the tracing system uses all routes in the platform file
1694 to re-create a "graph" of the platform and register it in the trace file.
1695 This option let the user tell the tracing system to use only the routes
1696 that are composed with just one link.
1701 It activates the categorized resource utilization tracing. It should
1702 be enabled if tracing categories are used by this simulator.
1705 tracing/uncategorized
1707 It activates the uncategorized resource utilization tracing. Use it if
1708 this simulator do not use tracing categories and resource use have to be
1714 A file with this name will be created to register the simulation. The file
1715 is in the Paje format and can be analyzed using Triva or Paje visualization
1716 tools. More information can be found in these webpages:
1717 <a href="http://triva.gforge.inria.fr/">http://triva.gforge.inria.fr/</a>
1718 <a href="http://paje.sourceforge.net/">http://paje.sourceforge.net/</a>
1723 This option only has effect if this simulator is SMPI-based. Traces the MPI
1724 interface and generates a trace that can be analyzed using Gantt-like
1725 visualizations. Every MPI function (implemented by SMPI) is transformed in a
1726 state, and point-to-point communications can be analyzed with arrows.
1731 This option only has effect if this simulator is SMPI-based. The processes
1732 are grouped by the hosts where they were executed.
1737 This option only has effect if this simulator is MSG-based. It traces the
1738 behavior of all categorized MSG tasks, grouping them by hosts.
1743 This option only has effect if this simulator is MSG-based. It traces the
1744 behavior of all categorized MSG processes, grouping them by hosts. This option
1745 can be used to track process location if this simulator has process migration.
1749 triva/categorized:graph_categorized.plist
1751 This option generates a graph configuration file for Triva considering
1752 categorized resource utilization.
1755 triva/uncategorized:graph_uncategorized.plist
1757 This option generates a graph configuration file for Triva considering
1758 uncategorized resource utilization.
1760 \subsubsection faq_tracing_example Example of Instrumentation
1762 A simplified example using the tracing mandatory functions.
1765 int main (int argc, char **argv)
1767 MSG_global_init (&argc, &argv);
1769 //(... after deployment ...)
1771 //note that category declaration must be called after MSG_create_environment
1772 TRACE_category_with_color ("request", "1 0 0");
1773 TRACE_category_with_color ("computation", "0.3 1 0.4");
1774 TRACE_category ("finalize");
1776 m_task_t req1 = MSG_task_create("1st_request_task", 10, 10, NULL);
1777 m_task_t req2 = MSG_task_create("2nd_request_task", 10, 10, NULL);
1778 m_task_t req3 = MSG_task_create("3rd_request_task", 10, 10, NULL);
1779 m_task_t req4 = MSG_task_create("4th_request_task", 10, 10, NULL);
1780 TRACE_msg_set_task_category (req1, "request");
1781 TRACE_msg_set_task_category (req2, "request");
1782 TRACE_msg_set_task_category (req3, "request");
1783 TRACE_msg_set_task_category (req4, "request");
1785 m_task_t comp = MSG_task_create ("comp_task", 100, 100, NULL);
1786 TRACE_msg_set_task_category (comp, "computation");
1788 m_task_t finalize = MSG_task_create ("finalize", 0, 0, NULL);
1789 TRACE_msg_set_task_category (finalize, "finalize");
1798 \subsubsection faq_tracing_analyzing Analyzing the SimGrid Traces
1800 The SimGrid library, during an instrumented simulation, creates a trace file in
1801 the Paje file format that contains the platform utilization for the simulation
1802 that was executed. The visualization analysis of this file is performed with the
1803 visualization tool <a href="http://triva.gforge.inria.fr">Triva</a>, with
1804 special configurations tunned to SimGrid needs. This part of the documentation
1805 explains how to configure and use Triva to analyse a SimGrid trace file.
1807 - <b>Installing Triva</b>: the tool is available in the INRIAGforge,
1808 at <a href="http://triva.gforge.inria.fr">http://triva.gforge.inria.fr</a>.
1809 Use the following command to get the sources, and then check the file
1810 <i>INSTALL</i>. This file contains instructions to install
1811 the tool's dependencies in a Ubuntu/Debian Linux. The tool can also
1812 be compiled in MacOSes natively, check <i>INSTALL.mac</i> file.
1814 $ svn checkout svn://scm.gforge.inria.fr/svn/triva
1819 - <b>Executing Triva</b>: a binary called <i>Triva</i> is available after the
1820 installation (you can execute it passing <em>--help</em> to check its
1821 options). If the triva binary is not available after following the
1822 installation instructions, you may want to execute the following command to
1823 initialize the GNUstep environment variables. We strongly recommend that you
1824 use the latest GNUstep packages, and not the packages available through apt-get
1825 in Ubuntu/Debian packaging systems. If you install GNUstep using the latest
1826 available packages, you can execute this command:
1828 $ source /usr/GNUstep/System/Library/Makefiles/GNUstep.sh
1830 You should be able to see this output after the installation of triva:
1832 $ ./Triva.app/Triva --help
1833 Usage: Triva [OPTIONS...] TRACE0 [TRACE1]
1834 Trace Analysis through Visualization
1837 --ti_frequency {double} Animation: frequency of updates
1838 --ti_hide Hide the TimeInterval window
1839 --ti_forward {double} Animation: value to move time-slice
1840 --ti_apply Apply the configuration
1841 --ti_update Update on slider change
1842 --ti_animate Start animation
1843 --ti_start {double} Start of time slice
1844 --ti_size {double} Size of time slice
1846 --comparison Compare Trace Files (Experimental)
1847 --graph Configurable Graph
1848 --list Print Trace Type Hierarchy
1849 --hierarchy Export Trace Type Hierarchy (dot)
1850 --stat Trace Statistics and Memory Utilization
1851 --instances List All Trace Entities
1852 --linkview Link View (Experimental)
1853 --treemap Squarified Treemap
1854 --merge Merge Trace Files (Experimental)
1855 --check Check Trace File Integrity
1857 --gc_conf {file} Graph Configuration in Property List Format
1858 --gc_apply Apply the configuration
1859 --gc_hide Hide the GraphConfiguration window
1861 Triva expects that the user choose one of the available options
1862 (currently <em>--graph</em> or <em>--treemap</em> for a visualization analysis)
1863 and the trace file from the simulation.
1865 - <b>Understanding Triva - time-slice</b>: the analysis of a trace file using
1866 the tool always takes into account the concept of the <em>time-slice</em>.
1867 This concept means that what is being visualized in the screen is always
1868 calculated considering a specific time frame, with its beggining and end
1869 timestamp. The time-slice is configured by the user and can be changed
1870 dynamically through the window called <em>Time Interval</em> that is opened
1871 whenever a trace file is being analyzed. The next figure depicts the time-slice
1872 configuration window.
1873 In the top of the window, in the space named <i>Trace Time</i>,
1874 the two fields show the beggining of the trace (which usually starts in 0) and
1875 the end (that depends on the time simulated by SimGrid). The middle of the
1876 window, in the square named <i>Time Slice Configuration</i>, contains the
1877 aspects related to the time-slice, including its <i>start</i> and its
1878 <i>size</i>. The gray rectangle in the bottom of this part indicates the
1879 <i>current time-slice</i> that is considered for the drawings. If the checkbox
1880 <i>Update Drawings on Sliders Change</i> is not selected, the button
1881 <i>Apply</i> must be clicked in order to inform triva that the
1882 new time-slice must be considered. The bottom part of the window, in the space
1883 indicated by the square <i>Time Slice Animation</i> can be used to advance
1884 the time-frame automatically. The user configures the amount of time that the
1885 time-frame will forward and how frequent this update will happen. Once this is
1886 configured, the user clicks the <i>Play</i> button in order to see the dynamic
1887 changes on the drawings.
1890 <a href="triva-time_interval.png" border=0><img src="triva-time_interval.png" width="50%" border=0></a>
1893 <b>Remarks:</b> when the trace has too many hosts or links, the computation to
1894 take into account a new time-slice can be expensive. When this happens, the
1895 <i>Frequency</i> parameter, but also updates caused by change on configurations
1896 when the checkbox <i>Update Drawings on Sliders
1897 Change</i> is selected will not be followed.
1899 - <b>Understanding Triva - graph</b>: this part of the documention explains how
1900 to analyze the traces using the graph view of Triva, when the user executes
1901 the tool passing <em>--graph</em> as parameter. Triva opens three windows when
1902 this parameter is used: the <i>Time Interval</i> window (previously described),
1903 the <i>Graph Representation</i> window, and the <em>Graph Configuration</em>
1904 window. The Graph Representation is the window where drawings take place.
1905 Initially, it is completely white waiting for a proper graph configuration input
1906 by the user. We start the description of this type of analysis by describing the
1907 <i>Graph Configuration</i> window (depicted below). By using a particular
1908 configuration, triva
1909 can be used to customize the graph drawing according to
1910 the SimGrid trace that was created with user-specific categories. Before delving
1911 into the details of this customization, let us first explain the major parts of
1912 the graph configuration window. The buttons located in the top-right corner can
1913 be used to delete, copy and create a new configuration. The checkbox in the
1914 top-middle part of the window indicates if the configuration typed in the
1915 textfield is syntactically correct (we are using the non-XML
1916 <a href="http://en.wikipedia.org/wiki/Property_list">Property List Format</a> to
1917 describe the configuration). The pop-up button located on the top-left corner
1918 indicates the selected configuration (the user can have multiple graph
1919 configurations). The bottom-left text field contains the name of the current
1920 configuration (updates on this field must be followed by typing enter on the
1921 keyboard to take into account the name change). The bottom-right <em>Apply</em>
1922 button activates the current configuration, resulting on an update on the graph
1926 <a href="triva-graph_configuration.png" border=0><img src="triva-graph_configuration.png" width="50%" border=0></a>
1929 <b>Basic SimGrid Configuration</b>: The figure shows in the big textfield the
1930 basic configuration that should be used during the analysis of a SimGrid trace
1931 file. The basic logic of the configuration is as follows:
1937 The nodes of the graph will be created based on the <i>node</i> parameter, which
1938 in this case is the different <em>"HOST"</em>s of the platform
1939 used to simulate. The <i>edge</i> parameter indicates that the edges of the
1940 graph will be created based on the <em>"LINK"</em>s of the platform. After the
1941 definition of these two parameters, the configuration must detail how
1942 <em>HOST</em>s and <em>LINK</em>s should be drawn. For that, the configuration
1943 must have an entry for each of the types used. For <em>HOST</em>, as basic
1944 configuration, we have:
1951 The parameter <em>size</em> indicates which variable from the trace file will be
1952 used to define the size of the node HOST in the visualization. If the simulation
1953 was executed with availability traces, the size of the nodes will be changed
1954 according to these traces. The parameter <em>scale</em> indicates if the value
1955 of the variable is <em>global</em> or <em>local</em>. If it is global, the value
1956 will be relative to the power of all other hosts, if it is local, the value will
1957 be relative locally.
1958 For <em>LINK</em> we have:
1968 For the types specified in the <em>edge</em> parameter (such as <em>LINK</em>),
1969 the configuration must contain two additional parameters: <em>src</em> and
1970 <em>dst</em> that are used to properly identify which nodes this edge is
1971 connecting. The values <em>source</em> and <em>destination</em> are always present
1972 in the SimGrid trace file and should not be changed in the configuration. The
1973 parameter <em>size</em> for the LINK, in this case, is configured as the
1974 variable <em>bandwidth</em>, with a <em>global</em> scale. The scale meaning
1975 here is exactly the same used for nodes. The last parameter is the GraphViz
1976 algorithm used to calculate the position of the nodes in the graph
1979 graphviz-algorithm = neato;
1982 <b>Customizing the Graph Representation</b>: triva is capable to handle
1983 a customized graph representation based on the variables present in the trace
1984 file. In the case of SimGrid, every time a category is created for tasks, two
1985 variables in the trace file are defined: one to indicate node utilization (how
1986 much power was used by that task category), and another to indicate link
1987 utilization (how much bandwidth was used by that category). For instance, if the
1988 user declares a category named <i>request</i>, there will be variables named
1989 <b>p</b><i>request</i> and a <b>b</b><i>request</i> (<b>p</b> for power and
1990 <b>b</b> for bandwidth). It is important to notice that the variable
1991 <i>prequest</i> in this case is only available for HOST, and
1992 <i>brequest</i> is only available for LINK. <b>Example</b>: suppose there are
1993 two categories for tasks: request and compute. To create a customized graph
1994 representation with a proportional separation of host and link utilization, use
1995 as configuration for HOST and LINK this:
2004 values = (prequest, pcomputation);
2017 values = (brequest, bcomputation);
2021 Where <i>sep_host</i> contains a composition of type <i>separation</i> where
2022 its max size is the <i>power</i> of the host and the variables <i>prequest</i>
2023 and <i>pcomputation</i> are drawn proportionally to the size of the HOST. And
2024 <i>sep_link</i> is also a separation where max is defined as the
2025 <i>bandwidth</i> of the link, and the variables <i>brequest</i> and
2026 <i>bcomputation</i> are drawn proportionally within a LINK.
2027 <i>This configuration enables the analysis of resource utilization by MSG tasks,
2028 and the identification of load-balancing issues, network bottlenecks, for
2030 <b>Other compositions</b>: besides <i>separation</i>, it is possible to use
2031 other types of compositions, such as gradients, and colors, like this:
2036 values = (numberOfTasks);
2040 values = (is_server);
2043 Where <i>gra_host</i> creates a gradient within a node of the graph, using a
2044 global scale and using as value a variable called <i>numberOfTasks</i>, that
2045 could be declared by the user using the optional tracing functions of SimGrid.
2046 If scale is global, the max and min value for the gradient will be equal to the
2047 max and min numberOfTasks among all hosts, and if scale is local, the max and
2048 min value based on the value of numberOfTasks locally in each host.
2049 And <i>color_host</i> composition draws a square based on a positive value of
2050 the variable <i>is_server</i>, that could also be defined by the user using the
2051 SimGrid tracing functions. \n
2052 <b>The Graph Visualization</b>: The next figure shows a graph visualization of a
2053 given time-slice of the masterslave_forwarder example (present in the SimGrid
2054 sources). The red color indicates tasks from the <i>compute</i> category. This
2055 visualization was generated with the following configuration:
2068 values = (pcompute, pfinalize);
2080 values = (bcompute, bfinalize);
2083 graphviz-algorithm = neato;
2088 <a href="triva-graph_visualization.png" border=0><img src="triva-graph_visualization.png" width="50%" border=0></a>
2092 - <b>Understading Triva - colors</b>: An important issue when using Triva is how
2093 to define colors. To do that, we have to know which variables are defined in
2094 the trace file generated by the SimGrid library. The parameter <em>--list</em>
2095 lists the variables for a given trace file:
2097 $ Triva -l masterslave_forwarder.trace
2115 We can see that HOST has seven variables (from power to pfinalize) and LINK has
2116 four (from bandwidth to bfinalize). To define a red color for the
2117 <i>pcompute</i> and <i>bcompute</i> (which are defined based on user category
2118 <i>compute</i>), execute:
2120 $ defaults write Triva 'pcompute Color' '1 0 0'
2121 $ defaults write Triva 'bcompute Color' '1 0 0'
2123 Where the three numbers in each line are the RGB color with values from 0 to 1.
2125 \subsection faq_modelchecking Model-Checking
2126 \subsubsection faq_modelchecking_howto How to use it
2127 To enable the experimental SimGrid model-checking support the program should
2128 be executed with the command line argument
2132 Properties are expressed as assertions using the function
2134 void MC_assert(int prop);
2137 \subsection faq_binding_lua Lua Binding
2138 Most of Simgrid modules require a good level in C programming, since simgrid is used to be as standard C library.
2139 Sometime users prefer using some kind of « easy scripts » or a language easier to code with, for their works,
2140 which avoid dealing with C errors, and sometime an important gain of time.
2141 Besides Java Binding, Lua and Ruby bindings are available since version 3.4 of Simgrid
2142 for MSG Module, and we are currenlty working on bindings for other modules.
2145 \subsubsection faq_binding_lua_about What is lua ?
2146 Lua is a lightweight, reflective, imperative and functional programming language,
2147 designed as a scripting language with extensible semantics as a primary goal (see official web site <a href="http://www.lua.org">here</a>).
2148 \subsubsection faq_binding_lua_why Why lua ?
2149 Lua is a fast, portable and powerful script language, quite simple to use for developpers.
2150 it combines procedural features with powerful data description facilities,
2151 by using a simple, yet powerful, mechanism of tables.
2152 Lua has a relatively simple C API compared to other scripting languages,
2153 and accordingly it provides a robust, easy to use it.
2154 \subsubsection faq_binding_lua_simgrid How to use lua in Simgrid ?
2155 Actually, the use of lua in Simgrid is quite simple, you have just to follow the same steps as coding with C in Simgird :
2156 - Coding functions coresponding to each process
2157 - loading the platforme/deployment XML file that describe the environment of simulation
2158 - and … Running the Simulation.
2160 \dontinclude lua/masterslave/master.lua
2161 \subsubsection faq_binding_lua_example_master_slave Master/Slave Example
2164 \until end_of_master
2165 we mainly use simgrid.Task.new(task_name,computation_size,communication_size) to create our MSG Task,
2166 then simgrid.Task.send(task,alias) to send it.
2167 we use also simgrid.Task.name(task), to get the task's name.
2169 \dontinclude lua/masterslave/slave.lua
2172 Here, we see the use of simgrid.Task.recv(alias) to receive a task with a specific alias,
2173 this function return directly the task recevied.
2175 \dontinclude lua/masterslave/master_slave.lua
2176 \li Set Environmenet and run application
2177 \until simgrid.clean()
2179 \subsubsection faq_binding_lua_example_data Exchanging Data
2180 You can also exchange data between Process using lua. for that, you have to deal with lua task as a table,
2181 since lua is based itself on a mechanism of tables,
2182 so you can exchange any kind of data (tables, matrix, strings,…) between process via tasks.
2186 task = simgrid.Task.new("data_task",task_comp,task_comm);
2187 task['matrix'] = my_matrix;
2188 task['table'] = my_table;
2189 task['message'] = "Hello from (Lua || Simgrid ) !! "
2191 simgrid.Task.send(task,alias)
2193 After creating task, we associate to it various kind of data with a specific key (string in this case)
2194 to distinguish between data variables. The receiver will use this key to access easily to datas.
2197 \li Receiver processe
2199 task = simgrid.Task.recv(alias);
2200 sender_matrix = task['matrix'];
2201 sender_table = task['table'];
2202 sender_message = task['message']
2205 Note that in lua, both sender and receiver share the same lua task.
2206 So that the receiver could joint data directly on the received task without sending it back.
2207 You can find a complet example (matrix multiplication case) in the file example/lua/mult_matrix.lua.
2210 \subsubsection faq_binding_lua_example_bypass Bypass XML
2211 maybe you wonder if there is a way to bypass the XML files,
2212 and describe your platform directly from the code, with lua bindings it's Possible !! how ?
2213 We provide some additional (tricky?) functions in lua that allows you to set up your own platform without using the XML files
2214 ( this can be useful for large platforms, so a simple for loop will avoid you to deal with an annoying XML File ;) )
2217 \li set Routing mode
2219 simgrid.AS.new{id="AS0",mode="Full"};
2224 simgrid.Host.new{id="Tremblay",power=98095000};
2225 simgrid.Host.new{id="Jupiter",power=76296000};
2226 simgrid.Host.new{id="Fafard",power=76296000};
2227 simgrid.Host.new{id="Ginette",power=48492000};
2228 simgrid.Host.new{id="Bourassa",power=48492000};
2230 we use simgrid.Host.new{id=id_host,power=power_host} to instanciate our hosts.
2235 simgrid.Link.new{id=i,bandwidth=252750+ i*768,latency=0.000270544+i*0.087}; -- some crazy values ;)
2238 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 ?)
2242 -- simgrid.Route.new(src_id,des_id,links_nb,links_list)
2243 simgrid.Route.new("Tremblay","Jupiter",1,{"1"});
2244 simgrid.Route.new("Tremblay","Fafard",6,{"0","1","2","3","4","8"});
2245 simgrid.Route.new("Tremblay","Ginette",3,{"3","4","5"});
2246 simgrid.Route.new("Tremblay","Bourassa",7,{"0","1","3","2","4","6","7"});
2248 simgrid.Route.new("Jupiter","Tremblay",1,{"1"});
2249 simgrid.Route.new("Jupiter","Fafard",7,{"0","1","2","3","4","8","9"});
2250 simgrid.Route.new("Jupiter","Ginette",4,{"3","4","5","9"});
2251 simgrid.Route.new("Jupiter","Bourassa",8,{"0","1","2","3","4","6","7","9"});
2254 for each host you have to specify which route to choose to access to the rest of hosts connected in the grid.
2258 simgrid.register_platform();
2260 Don't forget to register your platform, that SURF callbacks starts their work ;)
2264 simgrid.Host.setFunction("Tremblay","Master",4,{"20","550000000","1000000","4"});
2265 simgrid.Host.setFunction("Bourassa","Slave",1,{"0"});
2266 simgrid.Host.setFunction("Jupiter","Slave",1,{"1"});
2267 simgrid.Host.setFunction("Fafard","Slave",1,{"2"});
2268 simgrid.Host.setFunction("Ginette","Slave",1,{"3"});
2270 you don't need to use a deployment XML file, thanks to simgrid.Host.setFunction(host_id,function,args_number,args_list)
2271 you can associate functions for each host with arguments if needed .
2275 simgrid.register_application();
2277 Yes, Here too you have to resgiter your application before running the simulation.
2279 the full example is distributed in the file examples/lua/master_slave_bypass.lua
2281 \subsection faq_binding_ruby Ruby Binding
2284 \subsubsection faq_binding_ruby_simgrid Use Ruby in Simgrid
2285 Since v3.4, the use of <a href="http://ruby-lang.org">ruby</a> in simgrid is available for the MSG Module.
2286 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.
2288 \subsubsection faq_binding_ruby_example Master/Slave Ruby Application
2289 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,
2290 with a 'main' function that describe the behaviour of the process during the simulation.
2299 class Master < MSG::Process
2300 # main : that function that will be executed when running simulation
2302 def main(args) # args is an array containing arguments for function master
2305 MSG::info("args["+String(i)+"]="+args[i])
2308 raise "Master needs 3 arguments" if size < 3
2309 numberOfTask = Integer(args[0])
2310 taskComputeSize = Float(args[1])
2311 taskCommunicationSize = Float(args[2])
2312 slaveCount = Integer(args[3])
2314 # Creates and sends the tasks
2315 for i in 0..numberOfTask-1
2316 task = Task.new("Task_"+ i.to_s, taskComputeSize , taskCommunicationSize);
2317 mailbox = "slave " + (i%slaveCount).to_s
2318 MSG::info("Master Sending "+ task.name + " to " + mailbox + " with Comput Size " +
2321 MSG::info("Master Done Sending " + task.name + " to " + mailbox)
2324 # Sending Finalize MSG::Tasks
2325 MSG::info("Master: All tasks have been dispatched. Let's tell everybody the computation is over.")
2326 for i in 0..slaveCount-1
2327 mailbox = "slave " + i.to_s
2328 finalize_task = Task.new("finalize",0,0)
2329 finalize_task.send(mailbox)
2331 MSG::info("Master : Everything's Done")
2337 the class MSG::Task contains methods that allows the management of the native MSG tasks.
2338 in master ruby code we used :
2339 - <i>MSG::Task.new(task_name,compute_size,communication_size)</i> : to instanciate a new task.
2340 - <i>MSG::Task.send(mailbox)</i> : to send the task via a mailbox alias.
2341 - <i>MSG::Task.name</i> : to get the task's name.
2345 class Slave < MSG::Process
2348 mailbox = "slave " + args[0]
2349 for i in 0..args.size-1
2350 MSG::debug("args["+String(i)+"]="+args[i])
2354 MSG::info("Slave '"+ mailbox +"' waiting for new task");
2355 task = Task.receive(mailbox)
2356 if (task.name == "finalize")
2360 MSG::info("Slave '" + mailbox + "' done executing task "+ task.name + ".")
2362 MSG::info("I'm done, see you")
2366 to receive a task, we use the method <i>MSG::Task.receive(mailbox)</i> that return a MSG:Task object (received task).
2375 if (ARGV.length == 2)
2376 MSG.createEnvironment(ARGV[0])
2377 MSG.deployApplication(ARGV[1])
2381 MSG.createEnvironment("platform.xml")
2382 MSG.deployApplication("deploy.xml")
2385 puts "Simulation time : " + MSG.getClock.to_s
2389 - <i>MSG.createEnvironment(platform_file)</i> : set up the environment
2390 - <i>MSG.deployApplication(deployment_file)</i> : load the deployment file description.
2391 - <i>MSG.run</i> : run the simulation
2393 \subsubsection faq_binding_ruby_data Exchanging data
2394 ruby bindings provides two ways to exchange data between ruby processes.
2395 \li MSG::Task.join & MSG::Task.data <br/>
2397 the MSG::Task class contains 2 methods that allows a data exchange between 2 process.
2399 -<i>MSG::Task.join</i> : makes possible to join any kind of ruby data within a task.
2403 myTable <<1<<-2<<45<<67<<87<<76<<89<<56<<78<<3<<-4<<99
2404 # Creates and send Task With the Table inside
2405 task = MSG::Task.new("quicksort_task",taskComputeSize, taskCommunicationSize);
2410 -<i>MSG::Task.data</i> : to access to the data contained into the task.
2413 task = MSG::Task.receive(recv_mailbox.to_s)
2415 quicksort(table,0,table.size-1)
2418 you can find a complet example illustrating the use of those methods in file /example/ruby/Quicksort.rb
2422 another 'object-oriented' way to do it, is to make your own 'task' class that inherit from MSG::Task ,
2423 and contains data you want to deal with, the only 'tricky' thing is that "the initializer" method has no effect !
2425 the use of some getter/setter methods would be the simple way to manage your data :)
2427 class PingPongTask < MSG::Task
2428 # The initialize method has no effect
2438 you can find an example of use in file example/ruby/PingPong.rb
2440 \section faq_troubleshooting Troubleshooting
2442 \subsection faq_trouble_lib_compil SimGrid compilation and installation problems
2444 \subsubsection faq_trouble_lib_config cmake fails!
2446 We know only one reason for the configure to fail:
2448 - <b>You are using a broken build environment</b>\n
2449 If symptom is that the configury magic complains about gcc not being able to build
2450 executables, you are probably missing the libc6-dev package. Damn Ubuntu.
2452 If you experience other kind of issue, please get in touch with us. We are
2453 always interested in improving our portability to new systems.
2455 \subsubsection faq_trouble_distcheck Dude! "ctest" fails on my machine!
2457 Don't assume we never run this target, because we do. Check
2458 http://cdash.inria.fr/CDash/index.php?project=Simgrid (click on
2459 previous if there is no result for today: results are produced only by
2460 11am, French time) and
2461 https://buildd.debian.org/status/logs.php?pkg=simgrid if you don't believe us.
2463 If it's failing on your machine in a way not experienced by the
2464 autobuilders above, please drop us a mail on the mailing list so that
2465 we can check it out. Make sure to read \ref faq_bugrepport before you
2468 \subsection faq_trouble_compil User code compilation problems
2470 \subsubsection faq_trouble_err_logcat "gcc: _simgrid_this_log_category_does_not_exist__??? undeclared (first use in this function)"
2472 This is because you are using the log mecanism, but you didn't created
2473 any default category in this file. You should refer to \ref XBT_log
2474 for all the details, but you simply forgot to call one of
2475 XBT_LOG_NEW_DEFAULT_CATEGORY() or XBT_LOG_NEW_DEFAULT_SUBCATEGORY().
2477 \subsubsection faq_trouble_pthreadstatic "gcc: undefined reference to pthread_key_create"
2479 This indicates that one of the library SimGrid depends on (libpthread
2480 here) was missing on the linking command line. Dependencies of
2481 libsimgrid are expressed directly in the dynamic library, so it's
2482 quite impossible that you see this message when doing dynamic linking.
2484 If you compile your code statically (and if you use a pthread version
2485 of SimGrid -- see \ref faq_more_processes), you must absolutely
2486 specify <tt>-lpthread</tt> on the linker command line. As usual, this should
2487 come after <tt>-lsimgrid</tt> on this command line.
2489 \subsection faq_trouble_errors Runtime error messages
2491 \subsubsection faq_flexml_limit "surf_parse_lex: Assertion `next limit' failed."
2493 This is because your platform file is too big for the parser.
2495 Actually, the message comes directly from FleXML, the technology on top of
2496 which the parser is built. FleXML has the bad idea of fetching the whole
2497 document in memory before parsing it. And moreover, the memory buffer size
2498 must be determined at compilation time.
2500 We use a value which seems big enough for our need without bloating the
2501 simulators footprints. But of course your mileage may vary. In this case,
2502 just edit src/surf/surfxml.l modify the definition of
2503 FLEXML_BUFFERSTACKSIZE. E.g.
2506 #define FLEXML_BUFFERSTACKSIZE 1000000000
2509 Then recompile and everything should be fine, provided that your version of
2510 Flex is recent enough (>= 2.5.31). If not the compilation process should
2513 A while ago, we worked on FleXML to reduce a bit its memory consumption, but
2514 these issues remain. There is two things we should do:
2516 - use a dynamic buffer instead of a static one so that the only limit
2517 becomes your memory, not a stupid constant fixed at compilation time
2518 (maybe not so difficult).
2519 - change the parser so that it does not need to get the whole file in
2520 memory before parsing
2521 (seems quite difficult, but I'm a complete newbe wrt flex stuff).
2523 These are changes to FleXML itself, not SimGrid. But since we kinda hijacked
2524 the development of FleXML, I can grant you that any patches would be really
2525 welcome and quickly integrated.
2527 <b>Update:</b> A new version of FleXML (1.7) was released. Most of the work
2528 was done by William Dowling, who use it in his own work. The good point is
2529 that it now use a dynamic buffer, and that the memory usage was greatly
2530 improved. The downside is that William also changed some things internally,
2531 and it breaks the hack we devised to bypass the parser, as explained in
2532 \ref faq_flexml_bypassing. Indeed, this is not a classical usage of the
2533 parser, and Will didn't imagine that we may have used (and even documented)
2534 such a crude usage of FleXML. So, we now have to repair the bypassing
2535 functionality to use the lastest FleXML version and fix the memory usage in
2538 \subsubsection faq_trouble_gras_transport GRAS spits networking error messages
2540 Gras, on real platforms, naturally use regular sockets to communicate. They
2541 are deeply hidden in the gras abstraction, but when things go wrong, you may
2542 get some weird error messages. Here are some example, with the probable
2545 - <b>Transport endpoint is not connected</b>: several processes try to open
2546 a server socket on the same port number of the same machine. This is
2547 naturally bad and each process should pick its own port number for this.\n
2548 Maybe, you just have some processes remaining from a previous experiment
2550 Killing them may help, but again if you kill -KILL them, you'll have to
2551 wait for a while: they didn't close there sockets properly and the system
2552 needs a while to notice that this port is free again.
2554 - <b>Socket closed by remote side</b>: if the remote process is not
2555 supposed to close the socket at this point, it may be dead.
2557 - <b>Connection reset by peer</b>: I found this on Internet about this
2558 error. I think it's what's happening here, too:\n
2559 <i>This basically means that a network error occurred while the client was
2560 receiving data from the server. But what is really happening is that the
2561 server actually accepts the connection, processes the request, and sends
2562 a reply to the client. However, when the server closes the socket, the
2563 client believes that the connection has been terminated abnormally
2564 because the socket implementation sends a TCP reset segment telling the
2565 client to throw away the data and report an error.\n
2566 Sometimes, this problem is caused by not properly closing the
2567 input/output streams and the socket connection. Make sure you close the
2568 input/output streams and socket connection properly. If everything is
2569 closed properly, however, and the problem persists, you can work around
2570 it by adding a one-second sleep before closing the streams and the
2571 socket. This technique, however, is not reliable and may not work on all
2573 Since GRAS sockets are closed properly (repeat after me: there is no bug
2574 in GRAS), it is either that you are closing your sockets on server side
2575 before the client get a chance to read them (use gras_os_sleep() to delay
2576 the server), or the server died awfully before the client got the data.
2578 \subsubsection faq_trouble_errors_big_fat_warning I'm told that my XML files are too old.
2580 The format of the XML platform description files is sometimes
2581 improved. For example, we decided to change the units used in SimGrid
2582 from MBytes, MFlops and seconds to Bytes, Flops and seconds to ease
2583 people exchanging small messages. We also reworked the route
2584 descriptions to allow more compact descriptions.
2586 That is why the XML files are versionned using the 'version' attribute
2587 of the root tag. Currently, it should read:
2589 <platform version="2">
2592 If your files are too old, you can use the simgrid_update_xml.pl
2593 script which can be found in the tools directory of the archive.
2595 \subsection faq_trouble_valgrind Valgrind-related and other debugger issues
2597 If you don't, you really should use valgrind to debug your code, it's
2600 \subsubsection faq_trouble_vg_longjmp longjmp madness in valgrind
2602 This is when valgrind starts complaining about longjmp things, just like:
2604 \verbatim ==21434== Conditional jump or move depends on uninitialised value(s)
2605 ==21434== at 0x420DBE5: longjmp (longjmp.c:33)
2607 ==21434== Use of uninitialised value of size 4
2608 ==21434== at 0x420DC3A: __longjmp (__longjmp.S:48)
2611 This is the sign that you didn't used the exception mecanism well. Most
2612 probably, you have a <tt>return;</tt> somewhere within a <tt>TRY{}</tt>
2613 block. This is <b>evil</b>, and you must not do this. Did you read the section
2616 \subsubsection faq_trouble_vg_libc Valgrind spits tons of errors about backtraces!
2618 It may happen that valgrind, the memory debugger beloved by any decent C
2619 programmer, spits tons of warnings like the following :
2620 \verbatim ==8414== Conditional jump or move depends on uninitialised value(s)
2621 ==8414== at 0x400882D: (within /lib/ld-2.3.6.so)
2622 ==8414== by 0x414EDE9: (within /lib/tls/i686/cmov/libc-2.3.6.so)
2623 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
2624 ==8414== by 0x414F937: _dl_open (in /lib/tls/i686/cmov/libc-2.3.6.so)
2625 ==8414== by 0x4150F4C: (within /lib/tls/i686/cmov/libc-2.3.6.so)
2626 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
2627 ==8414== by 0x415102D: __libc_dlopen_mode (in /lib/tls/i686/cmov/libc-2.3.6.so)
2628 ==8414== by 0x412D6B9: backtrace (in /lib/tls/i686/cmov/libc-2.3.6.so)
2629 ==8414== by 0x8076446: xbt_dictelm_get_ext (dict_elm.c:714)
2630 ==8414== by 0x80764C1: xbt_dictelm_get (dict_elm.c:732)
2631 ==8414== by 0x8079010: xbt_cfg_register (config.c:208)
2632 ==8414== by 0x806821B: MSG_config (msg_config.c:42)
2635 This problem is somewhere in the libc when using the backtraces and there is
2636 very few things we can do ourselves to fix it. Instead, here is how to tell
2637 valgrind to ignore the error. Add the following to your ~/.valgrind.supp (or
2638 create this file on need). Make sure to change the obj line according to
2639 your personnal mileage (change 2.3.6 to the actual version you are using,
2640 which you can retrieve with a simple "ls /lib/ld*.so").
2643 name: Backtrace madness
2645 obj:/lib/ld-2.3.6.so
2650 fun:__libc_dlopen_mode
2653 Then, you have to specify valgrind to use this suppression file by passing
2654 the <tt>--suppressions=$HOME/.valgrind.supp</tt> option on the command line.
2655 You can also add the following to your ~/.bashrc so that it gets passed
2656 automatically. Actually, it passes a bit more options to valgrind, and this
2657 happen to be my personnal settings. Check the valgrind documentation for
2660 \verbatim export VALGRIND_OPTS="--leak-check=yes --leak-resolution=high --num-callers=40 --tool=memcheck --suppressions=$HOME/.valgrind.supp" \endverbatim
2662 \subsubsection faq_trouble_backtraces Truncated backtraces
2664 When debugging SimGrid, it's easier to pass the
2665 --disable-compiler-optimization flag to the configure if valgrind or
2666 gdb get fooled by the optimization done by the compiler. But you
2667 should remove these flag when everything works before going in
2668 production (before launching your 1252135 experiments), or everything
2669 will run only one half of the true SimGrid potential.
2671 \subsection faq_deadlock There is a deadlock in my code!!!
2673 Unfortunately, we cannot debug every code written in SimGrid. We
2674 furthermore believe that the framework provides ways enough
2675 information to debug such informations yourself. If the textual output
2676 is not enough, Make sure to check the \ref faq_visualization FAQ entry to see
2677 how to get a graphical one.
2679 Now, if you come up with a really simple example that deadlocks and
2680 you're absolutely convinced that it should not, you can ask on the
2681 list. Just be aware that you'll be severely punished if the mistake is
2682 on your side... We have plenty of FAQ entries to redact and new
2683 features to implement for the impenitents! ;)
2685 \subsection faq_surf_network_latency I get weird timings when I play with the latencies.
2687 OK, first of all, remember that units should be Bytes, Flops and
2688 Seconds. If you don't use such units, some SimGrid constants (e.g. the
2689 SG_TCP_CTE_GAMMA constant used in most network models) won't have the
2690 right unit and you'll end up with weird results.
2692 Here is what happens with a single transfer of size L on a link
2693 (bw,lat) when nothing else happens.
2696 0-----lat--------------------------------------------------t
2697 |-----|**** real_bw =min(bw,SG_TCP_CTE_GAMMA/(2*lat)) *****|
2700 In more complex situations, this min is the solution of a complex
2701 max-min linear system. Have a look
2702 <a href="http://lists.gforge.inria.fr/pipermail/simgrid-devel/2006-April/thread.html">here</a>
2703 and read the two threads "Bug in SURF?" and "Surf bug not
2704 fixed?". You'll have a few other examples of such computations. You
2705 can also read "A Network Model for Simulation of Grid Application" by
2706 Henri Casanova and Loris Marchal to have all the details. The fact
2707 that the real_bw is smaller than bw is easy to understand. The fact
2708 that real_bw is smaller than SG_TCP_CTE_GAMMA/(2*lat) is due to the
2709 window-based congestion mechanism of TCP. With TCP, you can't exploit
2710 your huge network capacity if you don't have a good round-trip-time
2711 because of the acks...
2713 Anyway, what you get is t=lat + L/min(bw,SG_TCP_CTE_GAMMA/(2*lat)).
2715 * if I you set (bw,lat)=(100 000 000, 0.00001), you get t = 1.00001 (you fully
2717 * if I you set (bw,lat)=(100 000 000, 0.0001), you get t = 1.0001 (you're on the
2719 * if I you set (bw,lat)=(100 000 000, 0.001), you get t = 10.001 (ouch!)
2721 This bound on the effective bandwidth of a flow is not the only thing
2722 that may make your result be unexpected. For example, two flows
2723 competing on a saturated link receive an amount of bandwidth inversely
2724 proportional to their round trip time.
2726 \subsection faq_bugrepport So I've found a bug in SimGrid. How to report it?
2728 We do our best to make sure to hammer away any bugs of SimGrid, but this is
2729 still an academic project so please be patient if/when you find bugs in it.
2730 If you do, the best solution is to drop an email either on the simgrid-user
2731 or the simgrid-devel mailing list and explain us about the issue. You can
2732 also decide to open a formal bug report using the
2733 <a href="https://gforge.inria.fr/tracker/?atid=165&group_id=12&func=browse">relevant
2734 interface</a>. You need to login on the server to get the ability to submit
2737 We will do our best to solve any problem repported, but you need to help us
2738 finding the issue. Just telling "it segfault" isn't enough. Telling "It
2739 segfaults when running the attached simulator" doesn't really help either.
2740 You may find the following article interesting to see how to repport
2741 informative bug repports:
2742 http://www.chiark.greenend.org.uk/~sgtatham/bugs.html (it is not SimGrid
2743 specific at all, but it's full of good advices).
2745 \author Arnaud Legrand (arnaud.legrand::imag.fr)
2746 \author Martin Quinson (martin.quinson::loria.fr)
2751 ******************************************************************
2752 * OLD CRUFT NOT USED ANYMORE *
2753 ******************************************************************
2756 \subsection faq_crosscompile Cross-compiling a Windows DLL of SimGrid from linux
2758 At the moment, we do not distribute Windows pre-compiled version of SimGrid
2759 because the support for this platform is still experimental. We know that
2760 some parts of the GRAS environment do not work, and we think that the others
2761 environments (MSG and SD) have good chances to work, but we didn't test
2762 ourselves. This section explains how we generate the SimGrid DLL so that you
2763 can build it for yourself. First of all, you need to have a version more
2764 recent than 3.1 (ie, a SVN version as time of writting).
2766 In order to cross-compile the package to windows from linux, you need to
2767 install mingw32 (minimalist gnu win32). On Debian, you can do so by
2768 installing the packages mingw32 (compiler), mingw32-binutils (linker and
2769 so), mingw32-runtime.
2771 You can use the VPATH support of configure to compile at the same time for
2772 linux and windows without dupplicating the source nor cleaning the tree
2773 between each. Just run bootstrap (if you use the SVN) to run the autotools.
2774 Then, create a linux and a win directories. Then, type:
2775 \verbatim cd linux; ../configure --srcdir=.. <usual configure flags>; make; cd ..
2776 cd win; ../configure --srcdir=.. --host=i586-mingw32msvc <flags>; make; cd ..
2778 The trick to VPATH builds is to call configure from another directory,
2779 passing it an extra --srcdir argument to tell it where all the sources are.
2780 It will understand you want to use VPATH. Then, the trick to cross-compile
2781 is simply to add a --host argument specifying the target you want to build
2782 for. The i586-mingw32msvc string is what you have to pass to use the mingw32
2783 environment as distributed in Debian.
2785 After that, you can run all make targets from both directories, and test
2786 easily that what you change for one arch does not break the other one.
2788 It is possible that this VPATH build thing breaks from time to time in the
2789 SVN since it's quite fragile, but it's granted to work in any released
2790 version. If you experience problems, drop us a mail.
2792 Another possible source of issue is that at the moment, building the
2793 examples request to use the gras_stub_generator tool, which is a compiled
2794 program, not a script. In cross-compilation, you need to cross-execute with
2795 wine for example, which is not really pleasant. We are working on this, but
2796 in the meanwhile, simply don't build the examples in cross-compilation
2797 (<tt>cd src</tt> before running make).
2799 Program (cross-)compiled with mingw32 do request an extra DLL at run-time to be
2800 usable. For example, if you want to test your build with wine, you should do
2801 the following to put this library where wine looks for DLLs.
2803 cp /usr/share/doc/mingw32-runtime/mingwm10.dll.gz ~/.wine/c/windows/system/
2804 gunzip ~/.wine/c/windows/system/mingwm10.dll.gz
2807 The DLL is built in src/.libs, and installed in the <i>prefix</i>/bin directory
2808 when you run make install.
2810 If you want to use it in a native project on windows, you need to use
2811 simgrid.dll and mingwm10.dll. For each DLL, you need to build .def file
2812 under linux (listing the defined symbols), and convert it into a .lib file
2813 under windows (specifying this in a way that windows compilers like). To
2814 generate the def files, run (under linux):
2815 \verbatim echo "LIBRARY libsimgrid-0.dll" > simgrid.def
2816 echo EXPORTS >> simgrid.def
2817 nm libsimgrid-0.dll | grep ' T _' | sed 's/.* T _//' >> simgrid.def
2818 nm libsimgrid-0.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> simgrid.def
2820 echo "LIBRARY mingwm10.dll" > mingwm10.def
2821 echo EXPORTS >> mingwm10.def
2822 nm mingwm10.dll | grep ' T _' | sed 's/.* T _//' >> mingwm10.def
2823 nm mingwm10.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> mingwm10.def
2826 To create the import .lib files, use the <tt>lib</tt> windows tool (from
2827 MSVC) the following way to produce simgrid.lib and mingwm10.lib
2828 \verbatim lib /def:simgrid.def
2829 lib /def:mingwm10.def
2832 If you happen to use Borland C Builder, the right command line is the
2833 following (note that you don't need any file.def to get this working).
2834 \verbatim implib simgrid.lib libsimgrid-0.dll
2835 implib mingwm10.lib mingwm10.dll
2838 Then, set the following parameters in Visual C++ 2005:
2839 Linker -> Input -> Additional dependencies = simgrid.lib mingwm10.lib
2841 Just in case you wonder how to generate a DLL from libtool in another
2842 project, we added -no-undefined to any lib*_la_LDFLAGS variables so that
2843 libtool accepts to generate a dynamic library under windows. Then, to make
2844 it true, we pass any dependencies (such as -lws2 under windows or -lpthread
2845 on need) on the linking line. Passing such deps is a good idea anyway so
2846 that they get noted in the library itself, avoiding the users to know about
2847 our dependencies and put them manually on their compilation line. Then we
2848 added the AC_LIBTOOL_WIN32_DLL macro just before AC_PROG_LIBTOOL in the
2849 configure.ac. It means that we exported any symbols which need to be.
2850 Nowadays, functions get automatically exported, so we don't need to load our
2851 header files with tons of __declspec(dllexport) cruft. We only need to do so
2852 for data, but there is no public data in SimGrid so we are good.