Other similar error
queue instance " " dropped because it is temporarily not available
queue instance " " dropped because it is disabled
Error display :
$qstat -j
queue instance "peteris.q@sge00" dropped because it is full
Explain: sge due to overload or discard all the queue is full
Solve the answer:
You seem to have reached a practical limit active job queue at any given time can handle the number. I'm not sure where SGE-defined maximum, but it seems likely:
max_jobs
Sun Grid Engine while allowing active (not completed) by the parameter control of the number of jobs. Defined limit value greater than 0. The default value of 0 means "unlimited." If the job is submitted beyond the max_jobs limit, the commit command will exit with an exit status 25, and displays the appropriate error message. Max_jobs changes will take effect immediately. This value is the only global configuration parameters. Execution host local configuration can not cover it.
From: HTTP : //gridscheduler.sourceforge.net/htmlman/htmlman5/sge_conf.html?pathrev=V62u5_TAG
If this is correct, then the value is infinite. However, SGE may be unable to properly manage approximately one million active work, so you may encounter this problem. I recommend that you use a job array , because this is the purpose of such operations, namely the management and operation of many nearly identical tasks .
SGE There are many online resources available for operation array, for example:
http://wiki.gridengine.info/wiki/index.php/Simple-Job-Array-Howto
http://talby.rcs.manchester.ac.uk/~ri/_linux_and_hpc_lib/sge_array.html
https://wiki.duke.edu/display/SCSC/SGE+Array+Jobs
If you edit a question depending on the specific requirements of each task, we will be happy to provide further assistance. For example, about one million tasks whether each requiring one or more parameters as input?
maxjobs
to0
solve this problem? Or, you have setmaxjobs
for the0
problem persists? - Vince 18 years at 12:53 on June 4Sun Grid Engine array job
1。
why?
Suppose you want to run a large number of jobs is largely the same: You may want to use a different parameter or parameters of multiple runs of the same program; or processing a thousand different input files. You can write a Perl script to generate all the necessary qsub file and write a BASH script to submit all these documents. However, this is not your time easy to use, Submit it would have on the cluster (login) node dire impact.
Use SGE array job will be better!
2。
what?
SGE array job can be described as a built-for-loop operation. This is a simple example:
By calculation, this is equivalent to 1000 submitted separate queue, wherein SGE_TASK_ID values 1,2,3. . 1000, where the input and output files indexed by ID. however:A slight change - in a separate directory (folder) to run each job:
3。
More
For more information on SGE job arrays, please visit:
4。
For more general circulation
SGE_TASK_ID do not have to start at 1; Increment not one. E.g:
#$ -t 100-995:5
So SGE_TASK_ID the value of 100,105,110,115 ... 995.
Incidentally, for example, the upper limit is not equal to an integral multiple of the case plus the increment,
#$ -t 1-42:6
SGE automatically changes the upper limit, that is,
5,
Related environmental variables
You can use the other three environment variables are automatically created, such as the following simple qsub shown script:
6。
Enter the file list
Can be sneaky - Suppose we have an input file list instead of the input file explicitly indexed by the suffix:
Source:
http://talby.rcs.manchester.ac.uk/~ri/_linux_and_hpc_lib/sge_array.html