Skip to content
GitLab
  1. 23 May, 2019 12 commits
  3. 22 May, 2019 5 commits
  5. 21 May, 2019 8 commits
  7. 17 May, 2019 2 commits
  9. 16 May, 2019 5 commits
    • Morris Jette's avatar
      Only allocate 1 CPU per node with the --overcommit option · dd7775ef
      Morris Jette authored 
      Previous select/cons_tres logic would allocate one CPU per task on the node

Bug 6981
      dd7775ef
    • Morris Jette's avatar
      modify task layout with --overcommit · 42d7e312
      Morris Jette authored 
      Modify task layout with --overcommit option plus a heterogeneous job
    allocation so that a cyclic task distribution can start happening before
    all CPUs on all nodes are fully allocated. The number of tasks per node
    will be unchanged from the previous algorithm, but tasks will be distributed
    in a cyclic fashion first and then extra tasks placed on nodes with more
    CPUs. Previously all CPUs would be fully allocated in a cyclic fashion,
    then excess tasks distributed evenly across all allocated nodes.
Bug 6981
      42d7e312
    • Dominik Bartkiewicz's avatar
      Store reservation flags in slurmdbd in a uint64_t. · 46d55dd4
      Dominik Bartkiewicz authored 
      Add warning to slurm.h.in that no new reservation flags can be
stored in slurmdbd in 19.05. (Although they could still be used by
slurmctld without issue.)

Note that the underlying RPC still uses uint32_t, but this will be
changed before 20.02 on master, and changing the column to uint32_t
in 19.05 just to change it again in 20.02 is best avoided.

Bug 6969.
      46d55dd4
    • Nathan Rini's avatar
      Fix memory leaks due to incomplete slurmdb_cluster_cond_t destructor. · 2038469f
      Nathan Rini authored 
      Free format_list, plugin_id_select_list, rpc_version_list in
 _free_cluster_cond_members().

Bug 7020.
      2038469f
    • Marshall Garey's avatar
      Fix archive loading events. · 0d0f9deb
      Marshall Garey authored 
      There was a syntax error in the mysql for inserting the event records
into the event table caused by commit 3d61b6aa. The syntax error was
a semicolon in the middle of the query, for example:

insert into "voyager_event_table" (time_start, time_end, node_name,
cluster_nodes, reason, reason_uid, state, tres) values ('1538669453',
'1539298628', 'v1', '', 'cold-start', '1017', '0',
'1=8,2=4000,5=8,1001=4,1002=1');, (<... another record>);, ...

Bug 7025.
      0d0f9deb
  11. 15 May, 2019 1 commit
    • Tim Wickberg's avatar
      Avoid call to slurm_get_slurmd_user_id() in _step_connect() if not slurmd. · 0a4c5234
      Tim Wickberg authored 
      For a stray socket, this call would cause nss_slurm to deadlock,
as any calling path that leads to slurm_conf_lock(), which will call
getpwuid(), which will re-enter the nss_slurm code, which will end up
back here but with the slurm_conf_lock already held, at which point
the process will never continue.

For nss_slurm, this means a node rebooting with stale sockets will hang
in the middle of the init process, which is a rather unpleasant experience.

So - only handle the stray socket cleanup within the slurmd process itself.

Bug 7030
      0a4c5234
  13. 13 May, 2019 1 commit
  15. 10 May, 2019 3 commits
    • Nate Rini's avatar
      Prevent leak of cluster_str in sacctmgr_list_runaway_jobs(). · bb9d5e79
      Nate Rini authored 
      Bug 6952.
      bb9d5e79
    • Marshall Garey's avatar
      Only archive 50k records at a time. · ddd49896
      Marshall Garey authored 
      Trying to archive too many records at once can result in archive files
that are too big to read or even too big to be written. Only archive 50k
records at a time, like we only purge 50k records at a time.

Bug 6033.
      ddd49896
    • Marshall Garey's avatar
      Handle duplicate archive file names. · 1e234c3d
      Marshall Garey authored 
      The time period of the archive file currently depends on submit or start
time and whether the purge period is in hours, days, or months.
Previously, if the archive file name already exists, we would overwrite
the old archive file with the assumption that these are duplicate
records being archived after an archive load. However, that could result
in lost records in a couple of ways:

  * If there were runaway jobs that were part of an old archive file's
  time period and are later fixed and then purged, the old file would
  be overwritten.
  * If jobs or steps are purged but there are still jobs or steps in
  that time period that are pending or running, the pending or running
  jobs and steps won't be purged. When they finish and are purged, the
  old file would be overwritten.

Instead of overwriting the old file, we append a number to the file name
to create a new file. This will also be important in an upcoming commit.

Bug 6033.
      1e234c3d
  17. 08 May, 2019 1 commit
  19. 07 May, 2019 2 commits