Opened 4 years ago

Closed 4 years ago

Last modified 4 years ago

#1569 closed defect (worksforme)

qsub delay

Reported by: Narsimha Owned by: Narsimha
Priority: normal Milestone:
Component: sge Version: 8.1.8
Severity: minor Keywords:
Cc: narsimha.418@…

Description

Hi all,

I am facing an issue with qsub command, when a script is issued with qsub as shown below:-

time qsub test.sh
Your job 12973 ("test.sh") has been submitted

real 0m59.665s
user 0m0.126s
sys 0m39.754s

It took 59 sec to generate the job id. Initially there is no delay. Since a week, we are facing this issue.

Kindly suggest how to resolve the issue.

Attachments (1)

signature.asc (819 bytes) - added by w.hay@… 4 years ago.
Added by email2trac

Download all attachments as: .zip

Change History (5)

comment:1 Changed 4 years ago by w.hay@…

On Tue, Mar 22, 2016 at 12:36:07PM +0000, SGE wrote:

#1569: qsub delay


Reporter: Narsimha | Owner: Narsimha

Type: defect | Status: new

Priority: normal | Milestone:

Component: sge | Version: 8.1.8

Severity: minor | Keywords:


Hi all,

I am facing an issue with qsub command, when a script is issued with qsub
as shown below:-

time qsub test.sh
Your job 12973 ("test.sh") has been submitted

real 0m59.665s
user 0m0.126s
sys 0m39.754s

It took 59 sec to generate the job id. Initially there is no delay. Since
a week, we are facing this issue.

Kindly suggest how to resolve the issue.

One thing that can cause slowdowns like this IIRC is if the filesystem the qmaster is using as a
spool is slow. This happens more easily if the spool is mounted on the qmaster via NFS. For this
reason I usually have the spool local to the qmaster and NFS exported to the rest of the cluster.

William

signature.asc

Changed 4 years ago by w.hay@…

Added by email2trac

comment:3 Changed 4 years ago by dlove

SGE <sge-bugs@…> writes:

One thing that can cause slowdowns like this IIRC is if the
filesystem the qmaster is using as a spool is slow. This happens
more easily if the spool is mounted on the qmaster via NFS. For this
reason I usually have the spool local to the qmaster and NFS exported
to the rest of the cluster.

The filesystem would be my initial guess too, but it needs debugging
locally. There's no evidence of an SGE bug anyhow.

You don't actually need to export qmaster's part of the spool, of
course, and there are reasons not to.

comment:4 Changed 4 years ago by dlove

  • Resolution set to worksforme
  • Status changed from new to closed

comment:5 Changed 4 years ago by Narsimha

Thank you for the reply.

I am having my spool directory located in the local disks on master and all compute nodes also.

And initially when sge is installed it used to work well without any delay but since a few days i am facing this issue.

Kindly help me to resolve this issue, also let me know why i am getting this delay.

Awaiting for your reply.

Note: See TracTickets for help on using tickets.