Difference between revisions of "Group Tools"
(3 intermediate revisions by the same user not shown) | |||
Line 5: | Line 5: | ||
All HPC groups are listed [[Unix groups|here]] and there are regular [[User Meetings]] to which every group is invited to send a representative. If you would like to add or change information regarding your group, please contact {{sc}} anytime. | All HPC groups are listed [[Unix groups|here]] and there are regular [[User Meetings]] to which every group is invited to send a representative. If you would like to add or change information regarding your group, please contact {{sc}} anytime. | ||
== Group Membership == | <!-- == Group Membership == | ||
The members of your group can be listed with the command | The members of your group can be listed with the command | ||
Line 12: | Line 12: | ||
If you see users in your group who are no longer a member, please contact {{sc}} to have the user removed. | If you see users in your group who are no longer a member, please contact {{sc}} to have the user removed. | ||
--> | |||
== Group Cluster Usage == | == Group Cluster Usage == | ||
The command <tt>report_group_usage</tt> can be used to find out how many core-hours have been accounted for your group. The data is updated on a monthly basis. At the moment, only core-hours are reported, in the future the share of your group in the cost of Scientific Computing will be estimated as well. | |||
The tool has a few options and a short description can be obtained with the <tt>--help</tt> option: | |||
<pre> | |||
$ report_group_usage --help | |||
Usage: | |||
report_group_usage [options] | |||
Report the group usage of the HPC cluster. | |||
Options: | |||
--version show program's version number and exit | |||
-h, --help show this help message and exit | |||
Available options: | |||
-m Show usage per month in addition to total | |||
-u Include usage per user in output. | |||
-g GROUP Specify group for report (admin users only). | |||
-y YEAR Specify year of report. | |||
</pre> | |||
''Example'': to get a report for the year 2018 including the usage per user, the following command can be used: | |||
<pre> | |||
$ report_group_usage -u -y 2018 | |||
======================================================================== | |||
= = | |||
= Usage Report for group hrz on cluster CARL = | |||
= ------------------------------------------ = | |||
= = | |||
= Year: 2018 Months: Mar-Dec = | |||
= = | |||
= used Core-h total: 78372891.8 (79.1%) = | |||
= CARL: 50355215.0 (89.7%) = | |||
= EDDY: 28017676.8 (65.1%) = | |||
= = | |||
= WallClock [h] (%) #Jobs = | |||
= ------------------------------------------------------------ = | |||
= hrz 3650.4 (0.0%) 860 = | |||
= |__ erle1100 61.0 26 = | |||
= |__ erma7540 1.3 6 = | |||
= |__ lees4820 3550.8 736 = | |||
= |__ lozi7895 35.7 88 = | |||
= |__ tugo5116 1.7 4 = | |||
= = | |||
======================================================================== | |||
</pre> | |||
The report shows you the name of the group (yours) and the associated cluster (CARL or EDDY) in the head line. In addition, the year and the available months are reported. The year can be changed with the option <tt>-y YEAR</tt>, all months will be reported for that year. The monthly data will be available at the beginning of the following month. | |||
The next three lines show the total number of core-hours for all users (in total and separately for CARL and EDDY) and the percentage with respect to the available core-hours (number of cores times hours reported). | |||
Finally, a table with the core-hours (column WallClock) used by your group is printed. The percentage is with repect to the total core-hours of the associated cluster. The last column shows the number of jobs. The details in the table depend on the options given. In the example, the option <tt>-u</tt> is specified, so the cpu-hours are shown in total for the group as well as per user. The option <tt>-m</tt> can be used to get a monthly listing (with or without user details). The default is to print only the total group usage. | |||
''Note'': Currently (Jan 15th), data is available for Mar-Dec, 2018, and Jan, Apr-Sep, 2017. If possible, data for Oct, 2017 to Feb, 2018 will be added soon. More recent data (from Jan, 2019) will be generated monthly. | |||
''Note'': A prediction of the share of the cost for Scientific Computing will be added later to the tool. The basic share (the part only depending on the core-hours of your group) can be already estimated easily with tool (and it is likely, that there will be no additional share). |
Latest revision as of 14:33, 11 February 2019
Managing your Group
On this page, we have documented tools you can use to manage your HPC group on the cluster. These tools are intended mainly for the head of a group but anyone in the group can use them. Therefore, the head of a group can delegate the management of the group to another member of his or her group.
All HPC groups are listed here and there are regular User Meetings to which every group is invited to send a representative. If you would like to add or change information regarding your group, please contact Scientific Computing anytime.
Group Cluster Usage
The command report_group_usage can be used to find out how many core-hours have been accounted for your group. The data is updated on a monthly basis. At the moment, only core-hours are reported, in the future the share of your group in the cost of Scientific Computing will be estimated as well.
The tool has a few options and a short description can be obtained with the --help option:
$ report_group_usage --help Usage: report_group_usage [options] Report the group usage of the HPC cluster. Options: --version show program's version number and exit -h, --help show this help message and exit Available options: -m Show usage per month in addition to total -u Include usage per user in output. -g GROUP Specify group for report (admin users only). -y YEAR Specify year of report.
Example: to get a report for the year 2018 including the usage per user, the following command can be used:
$ report_group_usage -u -y 2018 ======================================================================== = = = Usage Report for group hrz on cluster CARL = = ------------------------------------------ = = = = Year: 2018 Months: Mar-Dec = = = = used Core-h total: 78372891.8 (79.1%) = = CARL: 50355215.0 (89.7%) = = EDDY: 28017676.8 (65.1%) = = = = WallClock [h] (%) #Jobs = = ------------------------------------------------------------ = = hrz 3650.4 (0.0%) 860 = = |__ erle1100 61.0 26 = = |__ erma7540 1.3 6 = = |__ lees4820 3550.8 736 = = |__ lozi7895 35.7 88 = = |__ tugo5116 1.7 4 = = = ========================================================================
The report shows you the name of the group (yours) and the associated cluster (CARL or EDDY) in the head line. In addition, the year and the available months are reported. The year can be changed with the option -y YEAR, all months will be reported for that year. The monthly data will be available at the beginning of the following month.
The next three lines show the total number of core-hours for all users (in total and separately for CARL and EDDY) and the percentage with respect to the available core-hours (number of cores times hours reported).
Finally, a table with the core-hours (column WallClock) used by your group is printed. The percentage is with repect to the total core-hours of the associated cluster. The last column shows the number of jobs. The details in the table depend on the options given. In the example, the option -u is specified, so the cpu-hours are shown in total for the group as well as per user. The option -m can be used to get a monthly listing (with or without user details). The default is to print only the total group usage.
Note: Currently (Jan 15th), data is available for Mar-Dec, 2018, and Jan, Apr-Sep, 2017. If possible, data for Oct, 2017 to Feb, 2018 will be added soon. More recent data (from Jan, 2019) will be generated monthly. Note: A prediction of the share of the cost for Scientific Computing will be added later to the tool. The basic share (the part only depending on the core-hours of your group) can be already estimated easily with tool (and it is likely, that there will be no additional share).