Re: ClamAV 0.90 (was Re: Mailing List still Active)

mimedefangjohncom.net said:
> I went over to .90 yesterday and reverted back to .88.7
as .90 killed my
> machine. With .88.7 it runs around 20% cpu usage and a
load of .70 with
> clamav .90 it runs at 90% with a load of 6.0

I tried upgrading today and experienced a similar problem.
Clamd was using 90% 
(and more) CPU and not answering queries which made me run
out of slaves etc. 
etc.

OS is FreeBSD 6.2, mimedefang-2.61_1 (with the
--mbox-patch).

I am not doing any fancy in clamd.conf - only default values
except which user 
it runs as.


Any clues?


--Ingeborg
-- 
Ingeborg Østrem Hellemo  --  ingeborgcc.uit.no  (Univ. of
Tromsø, Norway)


_______________________________________________
NOTE: If there is a disclaimer or other legal boilerplate in
the above
message, it is NULL AND VOID.  You may ignore it.

Visit http://www.mimedefang.org and http://www.roaringpengu
in.com
MIMEDefang mailing list MIMEDefanglists.roaringpenguin.com
http://lists.roaringpenguin.com/mailman/listinfo/mime
defang

Hi,

>> I went over to .90 yesterday and reverted back to
.88.7 as .90 killed my
>> machine. With .88.7 it runs around 20% cpu usage
and a load of .70 with
>> clamav .90 it runs at 90% with a load of 6.0
>
> I tried upgrading today and experienced a similar
problem. Clamd was using 90%
> (and more) CPU and not answering queries which made me
run out of slaves etc.
> etc.
>
> OS is FreeBSD 6.2, mimedefang-2.61_1 (with the
--mbox-patch).
>
> I am not doing any fancy in clamd.conf - only default
values except which user
> it runs as.

Yes, I've seen similar problems on FreeBSD 6.2 with
lpthreads.

Use libmap.conf entries like those:

libpthread.so.2         libthr.so.2
libpthread.so           libthr.so
libc_r.so.5             libpthread.so.2

and only use a maximum of 20-30 threads. If you use more
threads libthr will have problems too ...

Martin
_______________________________________________
NOTE: If there is a disclaimer or other legal boilerplate in
the above
message, it is NULL AND VOID.  You may ignore it.

Visit http://www.mimedefang.org and http://www.roaringpengu
in.com
MIMEDefang mailing list MIMEDefanglists.roaringpenguin.com
http://lists.roaringpenguin.com/mailman/listinfo/mime
defang

Hi,

>> I tried upgrading today and experienced a similar
problem. Clamd was using 
>> 90%
>> (and more) CPU and not answering queries which made
me run out of slaves 
>> etc.

This patch did fix it for me:

--- clamav-0.90/shared/misc.c      Sun Feb 11 01:35:22 2007
+++ clamav-0.90/shared/misc.c  Tue Feb 20 23:55:57 2007
 -332,9
+332,7 
         dup2(i, 2);
      }

-    if(fork())
-       exit(0);
-
+    daemon(0,0);
      setsid();
  #endif
  }

Can you confirm this ?

Martin
_______________________________________________
NOTE: If there is a disclaimer or other legal boilerplate in
the above
message, it is NULL AND VOID.  You may ignore it.

Visit http://www.mimedefang.org and http://www.roaringpengu
in.com
MIMEDefang mailing list MIMEDefanglists.roaringpenguin.com
http://lists.roaringpenguin.com/mailman/listinfo/mime
defang

> This patch did fix it for me:
>

False alarm, daemon(2) still calls fork and makes clamd
spending CPU cycles.
I'm  using now a modified startup-script and call clamd with
daemon(8).

Martin
_______________________________________________
NOTE: If there is a disclaimer or other legal boilerplate in
the above
message, it is NULL AND VOID.  You may ignore it.

Visit http://www.mimedefang.org and http://www.roaringpengu
in.com
MIMEDefang mailing list MIMEDefanglists.roaringpenguin.com
http://lists.roaringpenguin.com/mailman/listinfo/mime
defang