Projinfo on Lindgren
The command projinfo on lindgren allows you to track how much time has been used in the time allocation(s) you are a member of. This is important as queue priority depends on how much time has been used in a time allocation compared to the quota, with jobs being blocked by the queue if the time allocation is too much over quota.
The simplest use of projinfo is just to do
This will give you information on the usage of all the time allocations you belong to one after another.
You can also use
to get help on the various options.
projinfo -p <project number>
will restrict the output to a group. Similarly you can get information on all the projects a specific user is a member of using
projinfo -u <username>
Information for project XXX-XX-X: PI_username: SNIC: TITLE Active from 20110701 to 20120707 Allocations: lindgren: 1200000.00 [corehours/month] 50000.00 [nodehours/month] 2083.33 [nodedays/month] target share 4.61% of 1508 nominally available nodes. share 7.03% of all enabled projects on lindgren. system: 29614.33 nodedays/month allocated out of 45240.00 moab: current 13.96% (target 7.03% CAP at 14.07%) lindgren: Completed jobs past 30 days: Total Average Average Username Group #jobs nodedays Percent #nodes q-days Full name -------- ----- ----- --------- ------- ------- ------- --------- TOTAL - 17122 42667.19 100.00 10.33 0.58 - GROUP XXX-XX-X 379 6161.04 14.44 28.04 0.39 - user1 XXX-XX-X 68 3005.99 7.05 48.00 0.49 user2 XXX-XX-X 121 2098.41 4.92 48.33 0.34 user3 XXX-XX-X 104 677.93 1.59 10.92 0.09 user4 XXX-XX-X 41 172.26 0.40 6.50 0.73 user5 XXX-XX-X 40 92.42 0.22 4.27 0.82 user6 XXX-XX-X 3 91.03 0.21 64.00 0.20 user7 XXX-XX-X 2 23.00 0.05 11.50 1.30
Hopefully the information is self explanatory, The header gives some information about the time allocation, including the size of the allocation in various units. The target share the percentage of the available time on the system jobs from this time allocation should be using if everyone used exactly 100% of their quota.
The system line tells you how many hours have been allocated to time allocations on the machine and the total amount of time that should be available.
The moab line shows the current usage as a percentage of all jobs that have been run together with the target share of all jobs for the allocation and what the fair share cap is (when usage is above this percentage then jobs from this time allocation will not be eligible to start.
Below that is a table detailing the usage by username of this allocation. TOTAL is the total of all users. GROUP is the total of this group. The time is given in node days, so to get core hours you need to multiply by 24*24 (i.e. 576).