Usage Statistics
Contents
Monthly Summaries 2013
Monthly Summaries 2012
Monthly Summaries 2011
Monthly Summaries 2010
Monthly Summaries 2009
Monthly Summaries 2008
qacct
You can compute statistics about your own or your project's CPU time with the "qacct" command. The numbers returned by qacct in the CPU column are in seconds instead of hours, so you'll need to divide by 3600 to make them comparable to the tables linked above.
For example, what was the usage recorded for project id "2009-1234" for the last 30 days?
> qacct -P 2009-1234 -d 30
Between 2008 Feb 1 and 2008 Feb 29, inclusive?
> qacct -P 2009-1234 -b 200802010000 -e 200802292359
Usage by members of John Doe's research group (Unix group jdoe) in the last 30 days, broken down by user?
> qacct -g jdoe -d 30 -o
Note that our monthly summaries are not strictly commensurable with qacct.
- we do not include WALLCLOCK, UTIME, STIME, IO and IOW
- we use different measures of memory consumption than qacct's MEMORY
- qacct does not report waiting times
- qacct includes jobs which started in a given period, we include jobs which ended in a given period
There are many options available for qacct.
| option | description |
|---|---|
| -b BeginTime | The earliest start time for jobs to be summarized, in the format [[CC]YY]MMDDhhmm[.SS]. |
| -d Days | The number of days to summarize and print accounting information on. If used together with the -b BeginTime option, jobs started within BeginTime to BeginTime + Days are counted. If used together with the -e EndTime option, count starts at EndTime - Days. If both are omitted, accounting period begins Days before now. |
| -e EndTime | The latest start time for jobs to be summarized, in the format [[CC]YY]MMDDhhmm[.SS]. |
| -g GroupName | The group name of the job owners to be included in the accounting. At ACENET, this is the same as the userid of the Principal Investigator. |
| -o Owner | The userid of the owner of the jobs for which accounting statistics are assembled. If the optional Owner argument is omitted, accounting data is listed for each relevant job owner separately. |
| -pe PEname | The name of the parallel environment for which usage is summarized. If PEname is omitted, accounting data is listed for each parallel environment separately. |
| -P Project | The name of the project for which usage is summarized. |
| -slots | Summarize accounting data by the number of queue slots each job occupied. |
Use "man qacct" for more details and further options.