Luke Youngblood wrote:

It's impossible to predict what will happen when applications run out of
memory and swap due to /tmp filling up.  This is what most people would call
a case of "operator error".  It might be useful to have a cfagent --cleanup
option that would cleanup any temp databases that might be out there, but I
don't think software can realistically be expected to do much of anything
once memory and swap has been exhausted.

Luke

-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On
Behalf Of Mark Burgess
Sent: Wednesday, May 25, 2005 6:14 AM
To: Yaroslav Halchenko
Cc: help-cfengine@gnu.org
Subject: Re: cfservd thrushes, nodes fail to get anything

I cannot see how this can occur. Any ideas from anyone else. There are
no loops. It is possible that this is internal to db.

Mark

On Sat, 2005-05-07 at 18:41 -0400, Yaroslav Halchenko wrote:
I've found the reason and probably that would be benefitial to adjust
cfservd to don't get into such situation again:

I had a leftover file /tmp/__db.testDATABASEcache

so strace revealed me infinite loop of
<snip>



_______________________________________________
Help-cfengine mailing list
Help-cfengine@gnu.org
http://lists.gnu.org/mailman/listinfo/help-cfengine
Perhaps it wouldn't be a trivial thing to have a simple perl script that monitors how full /tmp is and then call tmpwatch if it excedes a certain threshold. This of course is assuming that cfengine dosen't have a better way to deal with it.

shellcommands:
      ....
"/usr/bin/perl tmpstatus.pl" define=filling_up

      filling_up::
"/usr/sbin/tmpwatch 2 /tmp" # delete files in /tmp >= 2 hours old ....

   -- Dustin

--
Dustin Sorge HPC System Administrator Pittsburgh Supercomputing Center Carnegie Mellon University
4400 Fifth Avenue
Pittsburgh, PA 15213


_______________________________________________
Help-cfengine mailing list
Help-cfengine@gnu.org
http://lists.gnu.org/mailman/listinfo/help-cfengine

Reply via email to