On Thu, 2004-08-19 at 05:17, Lutz Petersen wrote: > Sure, but that's only a workaround. We have mailservers that > get clamd restartet (via clamdwatch.pl) nearly 10-15 times a day. > Every time clamd hangs that has consequences to the mail-flow, and > that's a real problem. Running clamd without softlimit ends in > clamd (after some time, some hours, some days or a week, I never > found out why) eating up all memory until the servers hook off. > So softlimit is a workaround, but not the solution. These memory > leaks may be the only real reason not to deal with clamav, in all > other relations this project is nice and fine. It would be very > positive to get the code 'de-leaked'. I'm not the C-programmer > to work on it, but if it helps our company could make some > donations to get clamd more (memory-) stable.
I run clamd for weeks at a time (until I choose to upgrade it) without any memory leaks. It appears people that report 'memory leaks' are running either Solaris or FreeBSD. It may be that there is a library on those systems that leaks memory. Until someone who can reproduce these memory leaks puts the effort in to find the cause, by using a memory bebugger, this issue is unlikely to get resolved. Needless to say, the developers don't see any leaks in our test systems, otherwise we would fix them. -trog
signature.asc
Description: This is a digitally signed message part
