Information for group managers

A group manager has the authority to manage the group course or the private cluster course.
By default, a group manager is set to an applicant of service course. Group manager can be added and changed via the user portal by the applicants of service courses.

A group manager can use the commands for managing the group. The dedicated commands of group manager allow group managers to manage queues and disks allocated to their groups and group members.

  • Before executing the command, change the current group using the newgrp command.

    $ id
    uid=12345(w12345) gid=12345(w12345) your_group=12345(w12345),19999(gr19999)
    $ newgrp - gr19999
    $ id
    uid=12345(w12345) gid=19999(gr19999) your_group=12345(w12345),19999(gr19999)

The current group manager can be found by using the group_managers command.

$ group_managers 
Group: gr19999, Managers: a59990,a59992

Group members can be checked and added by using the group_members command. Group members can submit jobs into the dedicated queue of their group, and can use the LARGE disk space.

Viewing the list of group members

$ group_members -g gr19999 -l
gr19999:a59990,a59991,a59992,a59993,a59994

Deleting the group member

$ group_members -g gr19999 -d b59999
deleting user b59999 from group gr19999

Adding the group member

$ group_members -g gr19999 -a b59999
adding user b59999 to group gr19999
  • Users managed by the payment manager of a service course are automatically registered as group members.
  • Members that are added by this command can submit jobs to the group queue and use the LARGE disk space.

When added a new member to a group, the new member cannot use the group queue at this time. After the changes are reflected in the job scheduler system, the new member will be allowed to use the group queue. The job scheduler system reflects the settings on the following morning after setting. If you need a faster reflection, please contact the programming support desk.

You can make backup settings of the LARGE disk space by using the group_backup command. LARGE disk space consists of the /LARGE0/groupname directory and the /LARGE1/groupname directory, one of the following status can be selected.

Status /LARGE0/groupname /LARGE1/groupname
Backup Safe(Make backup) Backup(Backup location)
Not Backup Unsafe(Not make backup) Unsafe(Not make backup)

The disk with the status of “Safe” or “Unsafe” can be used.

Checking the backup settings

$ group_backup  -l
 Num  Filesystem             Status  Filesystem             Status
  1)  /LARGE0/gr19999  ... Safe    /LARGE1/gr19999    ... Backup <- Means to make backup

Setting the status to “Not Backup”

$ group_backup  --unsafe 1
/LARGE0/gr19999: Safe   => UnSafe
/LARGE1/gr19999: Backup => UnSafe

Checking the changes

$ group_backup  -l
 Num  Filesystem             Status  Filesystem             Status
  1)  /LARGE0/gr19999    ... UnSafe  /LARGE1/gr19999    ... UnSafe  <- Means to make no backup

Setting the status to “Backup”

$ group_backup  --safe 1
/LARGE0/gr19999: Unsafe => Safe
/LARGE1/gr19999: Unsafe => Backup

The group_trash command allows its users to delete files(move to trash) in the LARGE disk space. It can delete the data files of users are no longer enrolled due to graduation, etc. You can recover the files that are deleted from the trash, but please note that the trash is emptied every Monday.

Deleting files by the group_trash command

$ group_trash  /LARGE0/gr19999/file1 
file1 to Trash (/LARGE0/gr19999/.DpcTrash/b59999/2009-04-10_1010)

You can view, add, and delete the members using queues, by using the queue_members command. Registered members can submit a job into the queue. By default, the proper group is registered with the group course queue.

Viewing the members using queues.

$ queue_members -q gr19999a -l
gr19999
w12345

Adding the members using queues.

$ queue_members -q gr19999a -a b67890

Deleting the members using queues.

$ queue_members -q gr19999a -d w12345
  • In every case, the -q option is required to specify the queue.
  • The job scheduler system reflects the changes including addition and removal of members on the following morning after changed.

You can set the job-scheduling policy of queues by an e-mail to the online consultation office. Both of the following policies can be set: The scheduling policy related to the execution order of jobs submitted to the queue(SCHEDULING_POLICY) and the passing policy related to passing on the pending jobs(PASSING_POLICY).

  • SCHEDULING_POLICY

    Settings Describes
    fcfs The submitted jobs are executed on a first-come, first-served basis(First-Come, First-Served). Default Setting.
    fairshare It calculates a dynamic priority for a fair-share of computing resources, and the jobs submitted by an user with a higher priority are executed in order of decreasing precedence.
  • PASSING_POLICY

    Settings Describes
    pass If the computing resources are adequate to run a particular job, the job are executed before jobs waiting to be executed. This utilizes the computing resources efficiently, but a large-scale job might never be executed. Default Setting.
    wait Even if the computing resources are enough, the passing between jobs is not implemented.
    backfill The passing is implemented in case of that the result of calculation based on running-time limit(-W) of each job does not affect the execution start time of other jobs. For example, a small-scale jobs can be completed while a large-scale job is waiting to execute, the computing resources can be utilized more efficiently.

You can cancel the submitted jobs using the group_qdel command.
Group managers can cancel any job in their queues, if the job is not their own.

$ group_qdel 123444
qdel: Job <123444> has finished


Copyright © Academic Center for Computing and Media Studies, Kyoto University, All Rights Reserved.