No locks available - full system table errors.

[Gino Castoldi_2]

Hi,

HP-UX 11.0 OVO 6.14 NNM 6.2 Oracle 8.7.1.3
L2000 4GB Memory.

We recently had a series of errors that generated these error messages in the OVO log file. "Can't lock queue file" xxx "due to full system table". "No locks available".

I noticed on the console numerous sendmail errors that were basically the same:

Nov 29 11:36:28 vpos1 sendmail[24285]: LAA24285: queueup: cannot lock tfLAA24285: Resource temporarily unavailable
Nov 29 11:36:29 vpos1 sendmail[24268]: LAA24268: queueup: cannot lock tfLAA24268: Error 0

The timestamps of the two types of error messages (OVO and sendmail) are also the same.

Can this problem be resolved by updating the
kernel parameters (nflocks, etc.)?

10 points to any good answer.
Thank you
Gino

[Robert-Jan Goossens · Accepted solution]

Gino,

Take a look at next doc.

http://www4.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000065450585

Regards,
Robert-Jan

[Jean-Louis Phelix]

Hi,

I'd rather look at nflocks, which should be around 10% of nfile (for 'normal' use), but which is hardcoded (no formula) to 200. You can check current utilization in glance (system tables). You could also get vmunix messages in the syslog file.

Regards.

[Gino Castoldi_2]

Hi,

Jean-Louis:
We do not have Glance. Is there another way I can check on those kernel parameters in real-time?

Robert-Jan:
I am having trouble accessing that doc.
Everytime I click on that URL it asks me to login even though I'm already logged in.
(I think it's a ITRC problem).
Can you post that document?

10 points to any good answer.
Thank you
Gino

[Robert-Jan Goossens]

http://www2.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000065450585

Document description: queueup: cannot lock
Document id: KBRC00003095

I'm not allowed to paste the doc inside the forum. Hope the above link to the US database will help.

Kind regards,
Robert-Jan

[Gino Castoldi_2]

Hi,

I read the document (KBRC00003095) and our
sendmail configuration should be ok.

We sent out around 4,000 emails within 10 minutes. The issue seems to be that we sent out too many messages and the system could not handle the number of file locks it needed.

10 points to any good answer.
Thank you
Gino

[Wessam Aly]

you can use "sar -v" to check the file system table

[Bernhard Mueller]

nflocks defines the maximum combined total number of file locks that are available system-wide to all processes at any given time.

Acceptable Values:
Minimum: 2
Maximum: Memory limited
Default: 200

kmtune -q nflocks

HTH
Regards
Bernhard