[ Skip to the content ]

Institute of Formal and Applied Linguistics Wiki

[ Back to the navigation ]

You are here: start » grid

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision 		Previous revision
Next revision Both sides next revision
grid [2017/09/27 20:45]
popel [Rules] 		grid [2017/10/02 17:28]
popel [Job monitoring]
Line 70: Line 70:
   # We have used two handy qsub parameters:   # We have used two handy qsub parameters:
   #  -cwd  ... the script is executed in the current directory (the default is your home)   #  -cwd  ... the script is executed in the current directory (the default is your home)
-  #  -j y  ... stdout and stderr outputs are merged and redirected to a file (''script.sh.o*'')+  #  -j y  ... stdout and stderr outputs are merged and redirected to a file (''script.sh.o$JOB_ID'')
   # We have also provided two parameters for our script "Hello" and "World".   # We have also provided two parameters for our script "Hello" and "World".
   # The qsub prints something like   # The qsub prints something like
Line 120: Line 120:
     * If your job needs more than one CPU (on a single machine) for most of the time, reserve the given number of CPU cores (and SGE slots) with <code>qsub -pe smp <number-of-CPU-cores></code> As you can see in [[#List of Machines]], the maximum is 32 cores. If your job needs e.g. up to 110% CPU most of the time and just occasionally 200%, it is OK to reserve just one core (so you don't waste). TODO: when using ''-pe smp -l mf=8G,amf=8G,h_vmem=12G'', which memory limits are per machine and which are per core?     * If your job needs more than one CPU (on a single machine) for most of the time, reserve the given number of CPU cores (and SGE slots) with <code>qsub -pe smp <number-of-CPU-cores></code> As you can see in [[#List of Machines]], the maximum is 32 cores. If your job needs e.g. up to 110% CPU most of the time and just occasionally 200%, it is OK to reserve just one core (so you don't waste). TODO: when using ''-pe smp -l mf=8G,amf=8G,h_vmem=12G'', which memory limits are per machine and which are per core?
     * If you are sure your job needs less than 1GB RAM, then you can skip this. Otherwise, if you need e.g. 8 GiB, you must always use ''qsub'' (or ''qrsh'') with ''-l mem_free=8G''. You should specify also ''act_mem_free'' with the same value and ''h_vmem'' with possibly a slightly bigger value. See [[#memory]] for details. TL;DR: <code>qsub -l mem_free=8G,act_mem_free=8G,h_vmem=12G</code>      * If you are sure your job needs less than 1GB RAM, then you can skip this. Otherwise, if you need e.g. 8 GiB, you must always use ''qsub'' (or ''qrsh'') with ''-l mem_free=8G''. You should specify also ''act_mem_free'' with the same value and ''h_vmem'' with possibly a slightly bigger value. See [[#memory]] for details. TL;DR: <code>qsub -l mem_free=8G,act_mem_free=8G,h_vmem=12G</code> 
-  * Be kind to your colleagues. If you are going to submit jobs that effectively occupy **more than one fifth of our cluster for more than several hours**, check if the cluster is free (with ''qstat -g c'' or ''qstat -u \*'') and/or ask your colleagues if they don't plan to use the cluster intensively in the near future. Note that if you allocate one slot (CPU core) on a machine, but (almost) all its RAM, you have effectively occupied the whole machine and all its cores. If you are submitting **more than 100 jobs**, consider using setting them a low priority (''-p -1024'', see below) or use [[#qunhold]]. Don't submit more than ca 2000 jobs at once (this can overload the SGE).+  * Be kind to your colleagues. If you are going to submit jobs that effectively occupy **more than one fifth of our cluster for more than several hours**, check if the cluster is free (with ''qstat -g c'' or ''qstat -u \*'') and/or ask your colleagues if they don't plan to use the cluster intensively in the near future. Note that if you allocate one slot (CPU core) on a machine, but (almost) all its RAM, you have effectively occupied the whole machine and all its cores. If you are submitting **more than 100 jobs**, consider using setting them a low priority (e.g. ''-p -1024'', see below) or use [[#qunhold]]. Don't submit more than ca 2000 jobs at once (this can overload the SGE).
  
  
Line 135: Line 135:
  
 ===== Advanced usage ===== ===== Advanced usage =====
 +
 +''qsub **-q** troja-all.q''
 +This way your job is submitted to the Troja queue. The default is ''ms-all.q''. You can also use e.g.
 +''-q '(troja*|ms*)''' to submit on any machine in those two queues (but **don't use ''-q '*'''** as this includes also [[:gpu|gpu.q]]),
 +''-q '*@hector[14]''' to submit on hecor1 or hector4,
 +''-q '[tm]*@!(hector*|iridium)''' to submit on any troja/ms machine except hectors and iridium.
 +However, usually you should specify just the queue (troja-all.q vs. ms-all.q), not a particular machine, and instead use ''-l'' to specify the needed resources in a general way.
 +
 +''qsub **-l** ...''
 +See ''man complex'' (run it on lrc or sol machines) for a list of possible resources you may require (in addition to ''mem_free'' etc. discussed above).
  
 ''qsub **-p** -100'' ''qsub **-p** -100''
 Define a priority of your job as a number between -1024 and 0. Only SGE admins may use a number higher than 0. The default is 0, i.e. the highest possible priority. SGE uses the priority to decide when to start which pending job in the queue (it computes a real number called ''prior'', which is reported in ''qstat'', which grows as the job is waiting in the queue). Note that once a job is started, you cannot "unschedule" it, so from that moment on, it is irrelevant what was its priority. Define a priority of your job as a number between -1024 and 0. Only SGE admins may use a number higher than 0. The default is 0, i.e. the highest possible priority. SGE uses the priority to decide when to start which pending job in the queue (it computes a real number called ''prior'', which is reported in ''qstat'', which grows as the job is waiting in the queue). Note that once a job is started, you cannot "unschedule" it, so from that moment on, it is irrelevant what was its priority.
 +
 ''qsub **-o** LOG.stdout **-e** LOG.stderr'' ''qsub **-o** LOG.stdout **-e** LOG.stderr''
-redirect std{out,err} to separate files with given names+redirect std{out,err} to separate files with given names, instead of the defaults ''$JOB_NAME.o$JOB_ID'' and ''$JOB_NAME.e$JOB_ID''.
  
 ''qsub **-@** optionfile'' ''qsub **-@** optionfile''
Line 174: Line 185:
 ''qsub **-soft** -l ... **-hard** -l ... -q ...'' ''qsub **-soft** -l ... **-hard** -l ... -q ...''
 By default, all the resource requirements (specified with ''-l'') and queue requirements (specified with ''-q'') are //hard//, i.e. your job won't be scheduled unless they can be fulfilled. You can use ''-soft'' to mark all following requirements as nice-to-have. And with ''-hard'' you can switch back to hard requirements. By default, all the resource requirements (specified with ''-l'') and queue requirements (specified with ''-q'') are //hard//, i.e. your job won't be scheduled unless they can be fulfilled. You can use ''-soft'' to mark all following requirements as nice-to-have. And with ''-hard'' you can switch back to hard requirements.
 +
 +''qsub **-sync** y''
 +This causes qsub to wait for the job to complete before exiting (with the same exit code as the job). Useful in scripts.
  
 ''**qalter**'' ''**qalter**''
 You can change some properties of already submitted jobs, which are still waiting in the queue (//pending//). You can change some properties of already submitted jobs, which are still waiting in the queue (//pending//).
  
-''**man** qsub qstat qhold queue_conf sge_types complex''+''**man** qsub qstat qalter qhold queue_conf sge_types complex''
 Find out all the gory details which are missing here. You'll have to do it one day anyway:-). Find out all the gory details which are missing here. You'll have to do it one day anyway:-).
  
Line 187: Line 201:
 === qunhold === === qunhold ===
 ''~stepanek/bin/qunhold'' tries to keep the number of running SGE jobs under a given threshold: all jobs over the threshold are held. If the number of running jobs goes below the threshold (default: 100), 10 jobs (by default) are unheld. Beware: if your jobs submit new jobs, you can get far over the threshold! ''~stepanek/bin/qunhold'' tries to keep the number of running SGE jobs under a given threshold: all jobs over the threshold are held. If the number of running jobs goes below the threshold (default: 100), 10 jobs (by default) are unheld. Beware: if your jobs submit new jobs, you can get far over the threshold!
 +
 +=== sshcwd ===
 +This is useful not only when sshing to sol machines. Add the following lines to your ''~/.bashrc''.
 +
 +<code>
 +function sshcwd () {
 +  # save the current history so that it is available
 +  # immediately on the remote machine
 +  history -a;
 +  # setup the working directory by setting WD
 +  ssh -X -Y -C -t $@ "WD='$PWD' /bin/bash --login -i";
 +}
 +
 +# use WD to setup the working directory
 +if [ -n "$WD" ]; then
 +  echo "Autochanging dir to $WD" >&2
 +  cd $WD;
 +fi
 +
 +alias sol1="sshcwd sol1.ufal.hide.ms.mff.cuni.cz"
 +</code>
  
 === In-script options === === In-script options ===
Line 202: Line 237:
 === Array jobs === === Array jobs ===
  
-If you have a set of tasks (of the same type) and want to run them on multiple machines,... +If you have a set of tasks (of the same type) and want to run them on multiple machines, use ''qsub -t''. 
-TODO +  * ''-t 1-n'' start array job with n jobs numbered 1 ... n 
-===== Monitorování úloh =====+  * environmental variable ''SGE_TASK_ID'' 
 +  * output and error files ''$JOB_NAME.[eo]$JOB_ID.$TASK_ID'' 
 +  * ''-t m-n[:s]'' start array job with jobs m, m + s, ..., n 
 +  * environmental variables ''SGE_TASK_FIRST, SGE_TASK_LAST, SGE_TASK_STEPSIZE'' 
 +  * ''-tc j'' run at most j jobs simultaneously 
 +  * ''-hold_jid_ad comma_separated_job_list'' array jobs that must finish before this job starts; task //i// of the current job depends only on task //i// of the specified jobs
  
-  * ''qstat [-u user]'' -- seznam úloh aktuálně běžících / ve frontě +===== Job monitoring =====
-  * ''qhost'' -- dostupné zdroje +
-  * ''/SGE/REPORTER/LRC-UFAL/bin/lrc_users_real_mem_usage -u user -w'' -- aktuální využití paměti uživatelem +
-  * ''/SGE/REPORTER/LRC-UFAL/bin/lrc_users_limits_requested -w'' -- nárokované požadavky uživatelů +
-  * ''/SGE/REPORTER/LRC-UFAL/bin/lrc_nodes_meminfo'' -- vypis vsech uzlu a stav vytiznosti pameti. +
-    * mem_total: celkova pamet uzlu +
-    * mem_free: tedy kolik je jeste volne pameti z pametove quoty uzlu +
-    * act_mem_free: kolik uzlu OPRAVDU zbyva volne pameti +
-    * mem_used: kolik je pameti skutecne pouzito +
-  * ''/SGE/REPORTER/LRC-UFAL/bin/lrc_state_overview'' -- celkový přehled o clusteru +
-    * celkovy pocet jader, pocet vyuzitych jader +
-    * celkova velikost RAM, kolik je ji fyzicky nepouzite, kolik je ji jeste nerezervovane +
-    * po jednotlivych uzivatelich (zrovna pocitajicich) -- kolik jim bezi uloh, kolik jich maji ve fronte a kolik z nich je ve stavu hold +
-  * ''cat /SGE/REPORTER/LRC-UFAL/stats/userlist.weight'' -- seznam uživatelů clusteru seřazený podle dosavadní aktivity (počet odeslaných úloh × čas, který běžely), aktualizovaný každý den v&nbsp;noci +
-  * [[https://ufaladm2.ufal.hide.ms.mff.cuni.cz/munin/ufal.hide.ms.mff.cuni.cz/lrc1.ufal.hide.ms.mff.cuni.cz/lrc_users.html|Munin: graf vytíženosti clusteru podle uživatelů]] (viditelný pouze ze sítě ÚFAL)+
  
-===== Časté a záludné problémy ===== +  * ''qstat [-u user]'' -- print a list of running/waiting jobs of a given user 
- +  * ''qhost'' -- print available/total resources 
- +  * ''/SGE/REPORTER/LRC-UFAL/bin/lrc_users_real_mem_usage -u user -w'' -- current memory usage of a given user 
-==== Submitnutý job může znovu submitovat ==== +  * ''/SGE/REPORTER/LRC-UFAL/bin/lrc_users_limits_requested -w'' -- required resources of all users 
- +  * ''/SGE/REPORTER/LRC-UFAL/bin/lrc_nodes_meminfo'' -- memory usage of all nodes 
-Danovy starší zkušenosti s clusterem PBS (nikoli SGE) říkaly, že tohle nejde. Ale jde to, aspoň u nás. Příkazy ''qsub'' a spol. jsou kromě hlavy clusteru dostupné i na všech strojích clusteru, samozřejmě pokud váš soubor ''.bashrc'', ''.cshrc'' apod. zajistí, že se i na nich provede inicializace prostředí SGE. +    * mem_total: 
- +    * mem_free: total memory minus reserved memory (using ''qsub -l mem_free'') for each node 
- +    * act_mem_free: really free memory 
- +    * mem_used: really used memory 
-==== Proměnné prostředí, nastavení vlastního prostředí ==== +  * ''/SGE/REPORTER/LRC-UFAL/bin/lrc_state_overview'' -- overall summary (with per-user stats for users with running jobs) 
- +  * ''cat /SGE/REPORTER/LRC-UFAL/stats/userlist.weight'' -- all users sorted according to their activity (number of submitted jobs × their average duration), updated each night 
-SGE spouští skripty v čistém prostředí. Nebuďte proto překvapeni, když vám skript na konzoli poběží dobře, ale po submitnutí fungovat nebude. Třeba nenašel potřebné programy v ''$PATH'' +  * [[https://ufaladm2.ufal.hide.ms.mff.cuni.cz/munin/ufal.hide.ms.mff.cuni.cz/lrc1.ufal.hide.ms.mff.cuni.cz/lrc_users.html|Munin: graph of cluster usage by day and user]] (accessible only from ÚFAL network, after accepting a security exception)
- +
-Zatím nevím přesně, které ze souborů ''.login'', ''.bashrc'' ap. SGE spouští, jestli vůbec nějaké. Naopak, experimentálně jsem ověřil, že ''qsub -S /bin/bash skript'' nenačte žádný z ''.bashrc'', ''.bash_profile'', ''.login'', ani ''.profile''. +
- +
-Z toho například také vyplývá, že bez ošetření se jako **Java** používá +
- +
-   java version "1.5.0" +
-   gij (GNU libgcj) version 4.1.2 20070502 (Red Hat 4.1.2-12) +
- +
-Pokud chcete submittovaný program pouštět ve svém oblíbeném prostředí (např. nastavení ''PATH''), musíte v obalujícím skriptu příslušné ''.bash*'' načíst. Vždy je ale bezpečnější všude psát plné cesty, než spoléhat na PATH. +
- +
-==== Jiný shell ==== +
- +
-Abych mohl poslat nějakou úlohu do fronty, musím pro ni vyrobit vlastní skript. Budiž, vyrobil jsem vlastní skript: +
- +
-<code> +
-#!/bin/bash +
-program > log.out 2> log.err +
-</code> +
- +
-Když tento skript spustím, stane se očekávané. Přesměrují se výstupy z daného programu do souborů a je to. +
- +
-Když takový skript submitnu, program se **nespustí**. V logu zjistím, že (standardní chybový) výstup shellu, který pouštěl můj skript praví kryptickou zprávu "Ambiguous redirect". +
- +
-Nebudu vás napínat, zde je vysvětlení: SGE ignoruje první řádek skriptu (ve skutečnosti je pravda horší, hledá v něm nějaké parametry pro sebe) a spouští skript v ''csh''. Rozdíl mezi bashem a csh se v primitivních skriptech na první pohled nepozná, pozná se až v konstrukci if-then-else, a také v přesměrovávání. csh nerozumí přesměrování ''2>'' +
- +
-Takto SGE přinutíte, aby použilo bash: +
- +
-<code> +
-  qsub -S /bin/bash skript +
-</code> +
- +
-Jinou možností je přesměrovat stderr a stdout pomocí syntaxe csh: +
- +
-<code> +
-( command >stdout_file ) >&stderr_file +
-</code> +
- +
- +
-==== bashrc a podobné nesmí nic vypisovat na konzoli ==== +
- +
-Opsáno z [[http://www.sara.nl/userinfo/lisa/usage/batch/index.html]]. +
- +
-It is important, that the files that are sourced during a login such as .bash_profile .profile .bashrc .login .cshrc don't produce any output when a non-interactive login is done. If they do, changes are that your job will run, but that the batch system is unable to deliver to you the standard output and error files. In that case the status of your job will be 'E' after the job is finished. Here is an example how you can test in a .bash_profile or .bashrc if this is an interactive login:+
  
 +===== Other =====
 +  * You can use environment variables ''$JOB_ID'', ''$JOB_NAME''.
 +  * One job can submit other jobs (but be careful with recursive:-)). A job submitted to the CPU cluster may submit GPU jobs (to the ''qpu.q'' queue).
 +  * It is important, that the files that are sourced during a login such as .bash_profile, .profile, .bashrc, .login etc. don't produce any output when a non-interactive login is done. If they do, changes are that your job will run, but that the batch system is unable to deliver to you the standard output and error files. In that case the status of your job will be 'E' after the job is finished. Here is an example how you can test in a .bash_profile or .bashrc if this is an interactive login:
 <code> <code>
 unset INTERACTIVE unset INTERACTIVE
Line 287: Line 274:
 fi fi
 </code> </code>
- +TODO: Is this restriction still true (for our cluster)? E.g. .bash_profile with /net/projects/SGE/user/sge_profile prints info messages on stderr and it is OK.
-==== Jak zjistit, jaké zdroje jsem pro svou úlohu požadoval ==== +
- +
-<code>qstat -j 973884,982737,984029,984030,984031,984034,984036 | grep resource +
-hard resource_list:         mem_free=50g +
-hard resource_list:         mem_free=200g +
-hard resource_list:         mem_free=16g +
-hard resource_list:         mem_free=16g +
-hard resource_list:         mem_free=16g +
-hard resource_list:         mem_free=31g</code> +
- +
-==== Jak rezervovat více jader na stejném stroji pro 1 job ==== +
- +
-<code> +
-qsub -pe smp <pocet jader> +
-</code>+
  

[ Back to the navigation ] [ Back to the content ]