On 5 Jul 2006 at 10:22, Stephen Gran wrote:

> On Wed, Jul 05, 2006 at 10:08:22AM +0100, Beginner said:
> > Hi,
> > 
> > Debian 3.1, Alpha GNU/Linux 2.4.27-2, clamd daemon 0.88.2 
> > 
> > I am using clamd with exim (exiscan) to scan emails. The other day the 
> > systems became un-
> > responsive and would not accept input from the console. It was getting the 
> > error "fork: Resource 
> > temporarily unavailable (35)".
> > 
> > The only reason I can find for this is the  large number of defunct clamd 
> > processes. I restart clamd 
> > yesterday and I already have 622

> 
> I have another report of problems on alpha (unaligned traps) - do you
> see anything in kern.log?  Can you debug this a bit (strace, gdb, etc)?

I just ran an strace on clamd until it spawned 2 defunct processes. 

clamav   29582  0.0  1.6 31016 21272 ?       Ss   11:11   0:00 /usr/sbin/clamd
clamav   29598  0.0  1.6 31016 21272 ?       S    11:11   0:00 /usr/sbin/clamd
clamav   29599  0.2  0.0     0    0 ?        Z    11:11   0:00 [clamd] <defunct>
clamav   29761  0.0  0.0     0    0 ?        Z    11:12   0:00 [clamd] <defunct>

The content of the output file from strace is a bit cryptic. I can't work out 
which of these pid is 
beind referred to so I am not sure what part of it to post. The output file has 
6115 lines in.

================== snip of strace output========
stat("/tmp/clamav-48bf4dd36561a84d", {st_mode=S_IFDIR|0700, st_size=4096, ...}) 
= 0
rmdir("/tmp/clamav-48bf4dd36561a84d")   = 0
close(6)                                = 0
close(5)                                = 0
munmap(0x2000001e000, 8192)             = 0
getdents64(4, /* 0 entries */, 8192)    = 0
close(4)                                = 0
gettimeofday({1152094304, 672942}, NULL) = 0
write(3, "Wed Jul  5 11:11:44 2006 -> Prot"..., 62) = 62
open("/dev/null", O_WRONLY)             = 4
close(0)                                = 0
dup2(4, 1)                              = 1
dup2(4, 2)                              = 2
chdir("/")                              = 0
fork()                                  = 29582
munmap(0x2000001c000, 8192)             = 0
exit_group(0)                           = ?
=========================================
 
> > The only potential cause for this that I can find is the ScanArchive 
> > directive in clamd.conf. I have 
> > tried to comment it out but when I re-start clamav the log shows "Archive 
> > support enabled"
> > 
> > Is there something I can do to stop these zombie processes from growing (a 
> > part from re-starting 
> > clamd)? Is there some reason why the ScanArcive option is still enabled 
> > when I have asked not 
> > specified it?
> 
> ScanArchive is one fo the default options, and in order to disable it,
> you need to add DisableDefaultScanOptions (I think I got the option
> right - check the man page), and then readd the defaults you do want.

Yes that is the option and it works thanx.

> Setting connection queue length to 15
> RECOMMENDED OPTIONS DISABLED.
> Archive support disabled.
> Portable Executable support enabled.
> Detection of broken executables enabled.

However it doesn't stop the defunct processes so I have enabled it again as 
it's not the source of 
my problem. Is there anything else I can do that might narrow things down?

Thanx.
Dp.


I
_______________________________________________
http://lurker.clamav.net/list/clamav-users.html

Reply via email to