Previous Thread
Next Thread
Print Thread
SYSTAT/K/Z Locks Ashell #34992 18 Feb 22 06:28 PM
Joined: Aug 2016
Posts: 329
J
John Andreasen Offline OP
Member
OP Offline
Member
J
Joined: Aug 2016
Posts: 329
Hi,

We see a chronic issue where all users on a system get locked when running SYSTAT/K/Z to kill zombie jobs. A message prints on the bottom status line that says something like waiting for access to unknown file.

Running these commands (I think Jack may have recommended them to Carl) on the server help in locating the offending job or jobs:
Code
[root@clearview-centos7 ~]# /vm/miame/bin/lslk | grep "ashell"
ashell     20945 8,16       292    15872  w 0  0  0 15871   0 /var/data/cvear/ashell/vm/miame/jobtbl.sys
[root@clearview-centos7 ~]# cp /vm/miame/jobtbl.sys /tmp/
[root@clearview-centos7 ~]# cp /vm/miame/qflock.sys /tmp/
[root@clearview-centos7 ~]# kill -9 20945
[root@clearview-centos7 ~]# /vm/miame/bin/lslk | grep "ashell"
ashell      2131 8,16       292    15872  w 0  0  0 15871   0 /var/data/cvear/ashell/vm/miame/jobtbl.sys
[root@clearview-centos7 ~]# kill -9 2131
[root@clearview-centos7 ~]# /vm/miame/bin/lslk | grep "ashell"

The A-Shell version on this particular customer is 6.5.1707.1 on CentOS 7.

I am not sure if anything can be done about this, but I did make a copy of the jobtbl.sys and qflock.sys before running the kill commands that I can send if that would help.

Thanks,
John Andreasen
Diversified Data Software

Re: SYSTAT/K/Z Locks Ashell [Re: John Andreasen] #34993 18 Feb 22 08:10 PM
Joined: Jun 2001
Posts: 11,645
J
Jack McGregor Online Content
Member
Online Content
Member
J
Joined: Jun 2001
Posts: 11,645
The problem here isn't so much the fault of the SYSTAT/K/Z but the apparent fact that certain jobs are somehow managing to lock the jobtbl and then get stuck with the lock set. (That will eventually cause a domino effect as jobs that need to scan the job table get hung up, and it will likely prevent new instances from starting.)

I don't think the jobtbl copy will help, but here's what probably will: rather than use kill -9, try using kill -TERM or kill -HUP, either of which should get logged in the ashlog, showing the program and location counter where the target job was when it received the signal. From there we can hopefully track down the cause. (If it doesn't succeed in terminating the target job, you can then resort to kill -9)

Re: SYSTAT/K/Z Locks Ashell [Re: John Andreasen] #34994 18 Feb 22 08:12 PM
Joined: Aug 2016
Posts: 329
J
John Andreasen Offline OP
Member
OP Offline
Member
J
Joined: Aug 2016
Posts: 329
OK, thanks Jack. I will try that the next time this happens.


Moderated by  Jack McGregor, Ty Griffin 

Powered by UBB.threads™ PHP Forum Software 7.7.3