Opened 2 years ago

Last modified 2 years ago

#982 assigned defect

PBS jobs never leave Q state

Reported by: paul.nord@… Owned by: skylar
Priority: major Milestone: 3.3.4
Component: Both Version: 3.3.4
Keywords: Cc:
Blocked By: Blocking:
Estimated Hours: 0 Total Hours: 0

Description

continuation of #981

Attachments (1)

test.pbs (94 bytes) - added by skylar 2 years ago.
test.pbs (Test PBS submit script)

Download all attachments as: .zip

Change History (8)

comment:1 Changed 2 years ago by skylar

  • Owner set to skylar
  • Status changed from new to assigned

this works:

sudo qrun job-id

comment:2 Changed 2 years ago by skylar

sudo qmgr -c 'unset queue batch resources_default.neednodes'

comment:3 Changed 2 years ago by skylar

in sched_prov/sched_config:

$pbs_server=node000.bccd.net

INVALID

Last edited 2 years ago by skylar (previous) (diff)

comment:4 Changed 2 years ago by skylar

WORKED!

sudo qmgr -c 'set queue batch started = True'

Last edited 2 years ago by skylar (previous) (diff)

comment:5 Changed 2 years ago by skylar

In 5797//cluster/svnroot:

fixing PBS setup re #982

comment:6 Changed 2 years ago by skylar

In 5799//cluster/svnroot:

enable scheduling re #982

comment:7 Changed 2 years ago by skylar

Testing:

  1. Run qmgr -c 'p s' and verify the following:
    1. set server scheduling = True
    2. set server default_queue = batch
    3. create queue batch
      1. set queue batch queue_type = Execution
      2. set queue batch enabled = True
      3. set queue batch started = True
  2. Start at least one diskless node
  3. Submit attached test.pbs enough times to fill all the slots on both the head node and diskless node

Changed 2 years ago by skylar

test.pbs (Test PBS submit script)

Note: See TracTickets for help on using tickets.