Clamd died on two of our servers within hours of each
other. Both of them died after doing a database self
check. Both of them were running fine before the check.
Excerpts below from clamd.log:
Sat Jul 16 04:56:29 2005 -> SelfCheck: Database status OK.
Sat Jul 16 04:57:20 2005 ->
/var/spool/qmailscan/tmp/a.zianet.com112151144050827376/1121511440.27384-0.a.zianet.com:
Unable to allocate memory ERROR
Sat Jul 16 01:16:03 2005 -> SelfCheck: Database status OK.
Sat Jul 16 01:16:48 2005 ->
/var/spool/qmailscan/tmp/b.zianet.com112149820850888076/1121498208.88083-0.b.zianet.com:
Unable to allocate memory ERROR
These are on FreeBSD 5.3 machines running clamav 0.86.
The error messages continue for about a minute then clamd
bombs out. Both of these clamd processes were started at
about the same time on each box and strangely enough they
died at about the same time. These boxes did *not* run
out of memory. Now whether they hit a soft limit I can't
say because I don't know how much memory the clamd process
was using at the time.
On some of our other machines clamd is using in upwards
of 115M. I have no idea why it needs that much memory
when on initial startup it takes all of 12-13M. Is there
a way to stop the self checking and loading of new definitions?
I would like, as a test, to start clamd on two different boxes
at the same time. One without the self checking and with. Then
let them run for 24 or so hours and take a look at memory usage.
Or has someone run into this before and there is a quick simple
fix? Or anyone have a better idea to try to narrow down the
problem?
_______________________________________________
http://lurker.clamav.net/list/clamav-users.html
