Différences
Ci-dessous, les différences entre deux révisions de la page.
| Les deux révisions précédentesRévision précédenteProchaine révision | Révision précédenteDernière révisionLes deux révisions suivantes | ||
| en:documentation:tools:sge [2018/09/04 09:21] – [GridEngine : Soumettre un job] cpetit | en:documentation:tools:sge [2020/08/25 15:58] – modification externe 127.0.0.1 | ||
|---|---|---|---|
| Ligne 1: | Ligne 1: | ||
| ====== GridEngine ====== | ====== GridEngine ====== | ||
| - | <WRAP center round todo 60%> | ||
| - | Under contruction... | ||
| - | </ | ||
| The job scheduler (or batch-queuing system) used in PSMN cluster is SGE, | The job scheduler (or batch-queuing system) used in PSMN cluster is SGE, | ||
| Ligne 14: | Ligne 11: | ||
| * the **queues for job submission** are described [[documentation: | * the **queues for job submission** are described [[documentation: | ||
| {{ : | {{ : | ||
| - | ==== Utilisation optimale des ressources | + | ==== Optimum use of resources |
| - | Pour utiliser au mieux les ressources, il convient de bien remplir les serveurs. Pour cela, nous disposons de deux moyens | + | To make the best use of resources, it is important to fill up the servers. For this, they are two strategies: |
| - | * remplir "au mieux" (best effort) , | ||
| - | * remplir par multiple de n coeurs (ou n est le nombre de coeurs physiques par serveur). | ||
| - | Remplir au mieux mène rapidement à un morcellement, entre les serveurs, des applications parallélisées via mpi. | + | * fill in "at best" (a best effort strategy), |
| + | * fill in with multiple of n cores (where n is the number of physical cores per server). | ||
| - | Le remplissage au mieux n'est donc mis en place pour les applications | + | Filling at best quickly leads to a fragmentation between the servers, of parallel |
| - | ==== Priorités ==== | + | Filling at best is therefore only implemented on a few queues for parallel applications. On other queues, the fill in with the multiple of n cores of an entire server is used. |
| - | Une priorité : | + | ==== Priorities ==== |
| - | * inversement proportionnelle au temps de calcul déjà consommé, | + | Job priority is: |
| - | * proportionnelle au temps d' | + | |
| - | est appliqué par GridEngine. Ceci afin de répartir plus équitablement les ressources disponibles. | + | * inversely proportional to the calculation time already consumed, |
| + | * proportional to waiting time and number of hearts requested. | ||
| + | |||
| + | This is to distribute the available resources more equitably. | ||
| ===== GridEngine : Submitting jobs ===== | ===== GridEngine : Submitting jobs ===== | ||
| - | The normal way to submit jobs to the cluster is using the '' | + | < |
| - | + | qsub programme <input >output | |
| - | < | + | |
| - | qsub myscript.sh | + | |
| </ | </ | ||
| - | The many options to the '' | ||
| - | |||
| - | For example a more complex submission: | ||
| <code bash> | <code bash> | ||
| - | qsub -V -m b -m e -e / | + | qsub -V -e / |
| -V : export environment variables | -V : export environment variables | ||
| - | -m b : mail @begin | ||
| - | -m e : mail @end | ||
| -e : where to put error files | -e : where to put error files | ||
| -o : where to put output files | -o : where to put output files | ||
| Ligne 56: | Ligne 47: | ||
| </ | </ | ||
| - | ** Nevertheless, | + | **It is simpler |
| - | ==== Comment choisir les files d'attente, de test ou de productions, | + | <note important> |
| - | De par les achats successifs de nœuds de calcul avec des architectures de génération différente, il n' | + | Voir [[documentation: |
| - | Concrètement le choix de la file d' | + | ==== How to choose the adapted queues for my needs? ==== |
| - | * si le critère principal est la rapidité de l' | + | Due to successive purchases of compute nodes with cores/CPU architectures of different generation, it was not possible to define a single queue. It is better to have different queues for each architecture, in order to achieve interesting performance for each queue. |
| - | * si le critère principal est le grand nombre de ressources (p.ex. un job avec beaucoup de cœurs, un job avec beaucoup de mémoire RAM), alors il faut plutôt s' | + | |
| - | Évidemment, | ||
| - | Et bien sur, pour la mise au point du code, il faut choisir une file d' | + | In concrete terms, the choice of the " |
| - | ===== GridEngine | + | |
| - | ==== Surveiller les jobs ==== | + | * if the main criterion is the speed of execution, you must look at what are the queues available to accept the job. The use of commands of the type '' |
| + | * if the main criterion is the large number of resources (//eg// a job with a lot of cores, a job with a lot of RAM,etc), then you have to move towards the queues that have a large number of resources (at least the resources requested by the job), even if waiting time in the queue is greater. | ||
| - | * Voir tous les jobs d'un utilisateur | + | Obviously, the above command (''qstat -g c'' |
| + | |||
| + | And, of course, when tuning ypur code, you have to choose a test queue that is closest to the intended " | ||
| + | ===== GridEngine : other useful commands ===== | ||
| + | |||
| + | ==== Checking job status ==== | ||
| + | |||
| + | * display job status of a specific user: | ||
| < | < | ||
| - | * Voir l' | + | * display queues status |
| < | < | ||
| - | * Voir tous les jobs de tous les utilisateurs en exécution (running) : | + | * display nodes status in a given queue: |
| + | |||
| + | < | ||
| + | |||
| + | * display the running | ||
| < | < | ||
| - | * Voir tous les jobs de tous les utilisateurs en attente (pending | + | * display the pending |
| < | < | ||
| - | * < | ||
| - | |||
| - | < | ||
| - | |||
| - | * < | ||
| - | |||
| - | < | ||
| - | |||
| - | * Voir l' | ||
| - | < | + | * display the status of a job in progress: |
| - | * Voir l' | + | < |
| - | < | + | * display the status of a job in progress with more details (longer): |
| - | * Voir l' | + | < |
| - | < | + | * display the status of a job in progress with even more details (even longer): |
| - | * <del>Enlever (clear) l' | + | <code>qstat -j < |
| - | < | ||
| - | * Voir les informations d' | + | * display information on a job afetr its completion |
| - | < | + | < |
| - | * Supprimer un job : | + | * delete a job: |
| - | < | + | < |
| - | * Supprimer un job (forcer) : | + | * delete a job (force deletion) : |
| - | < | + | < |
| ==== Accounting ==== | ==== Accounting ==== | ||
| - | <note important> | + | <note important> |
| - | * Détails des jobs sur les 30 derniers jours : | + | * Job details for the last 30 days: |
| < | < | ||
| - | * Consommation d' | + | * CPU hours consumption |
| < | < | ||
| Ligne 143: | Ligne 133: | ||
| < | < | ||
| - | * Consommation d' | + | * CPU hours consumption(utime |
| < | < | ||
| - | ==== Jobs qui ont des problèmes | + | ==== Troubleshootings: ==== |
| - | Lancer la commande suivante | + | Run the command: |
| < | < | ||
| - | et regarder les deux dernières colonnes | + | on the output, look at the last two columns: |
| - | * aoACD : Nombre de coeurs qui sont au moins dans un des états suivants | + | |
| - | * a Load threshold alarm | + | |
| - | * o Orphaned | + | |
| - | * A Suspend threshold alarm | + | |
| - | * C Suspended by calendar | + | |
| - | * D Disabled by calendar | + | |
| - | * cdsuE : Nombre de coeurs qui sont au moins dans un des états suivants | + | |
| - | * c Configuration ambiguous | + | |
| - | * d Disabled | + | |
| - | * s Suspended | + | |
| - | * u Unknown | + | |
| - | * E Error | + | |
| - | ==== Status (etats) de jobs possibles | + | ==== Possible job status: ==== |
| - | * d(eletion), | + | |
| - | * E(rror), | + | |
| - | * h(old), | + | |
| - | * r(unning), | + | |
| - | * R(estarted), | + | |
| - | * s(uspended), | + | |
| - | * S(uspended), | + | |
| - | * t(ransfering), | + | |
| - | * T(hreshold), | + | |
| - | * w(aiting). | + | |
| - | ===== GridEngine : Variables d' | + | ===== GridEngine: |
| - | <note important>''# | + | <note important>''# |
| - | * SGE_O_WORKDIR : répertoire d'où à été soumis le job, utilisable dans les scripts | + | * SGE_O_WORKDIR : directory where the job was submited, re-usable in scripts |
| - | * NSLOTS : nombre de coeurs demandé | + | * NSLOTS : number of slots/cores requested |
| - | * JOB_ID : ID du job (unique) | + | * JOB_ID : job id (unique) |
| - | * JOB_NAME : nom du job (-N) | + | * JOB_NAME : name of the job (-N) |
| - | * PE_HOSTFILE : fichier de hosts | + | * PE_HOSTFILE : hosts files (for MPI jobs) |
| - | ===== Références | + | ===== References |
| * http:// | * http:// | ||
| Ligne 205: | Ligne 195: | ||
| * http:// | * http:// | ||
| * https:// | * https:// | ||
| - | * http:// | + | * http:// |