I wrote a short perl script to send a command to clamd's Unix socket to
force 0.90.1 clamd to reload:
#!/usr/local/bin/perl
use IO::Socket::UNIX;
$clamSocket = "/var/spool/clamav/clamd.socket";
$timeout = 60;
$mysock = new IO::Socket::UNIX(Type => SOCK_STREAM,
Timeout => $timeout,
Peer => $clamSocket );
$mysock->send("RELOAD");
And I notice that the cpu usage approaches 90% for nearly 60 seconds -
ramping up to max in about 10 seconds, holding for about 30 seconds,
then backing down. It's reading the databases, of course, but it seems
to not do large block reads, rather it grabs one record per fetch.
That's a lot of little reads, and IO Wait goes up quite a bit, too. I
don't recall 0.88.x doing this. Is it new? Or am I not following truss
output correctly?
Sun Sparc Netra X2, 400mhz, 2g RAM, Sol8, whipped, tired, but won't die.
dp
_______________________________________________
Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net
http://lurker.clamav.net/list/clamav-users.html
