bruce
06-06-2006, 18:00
We have been successfully running teamspeak on a Linux-Vserver Virtual Server for some time. We just moved to a new server (dual p4 2.8 with kernel 2.6.16.14 SMP), using the same 'vps image' as before but TS locks up.
How it happens?
TS starts sucessfully. People can connect to it. After some time everyone times out, the connection ont he tcp port is "accepted" but not answered. A 'ps' shows all processes are active and seem to be ok. This is not a network issue since a restart on ts makes it work again.
This is happening a lot, like once an hour.
An strace on the active processes repeats this:
=================================
select(23, [22], [], [], {0, 100000}[{fd=13, events=POLLIN}], 1, 52) = 0
gettimeofday({1149605956, 472366}, NULL) = 0
gettimeofday({1149605956, 472414}, NULL) = 0
poll() = 0 (Timeout)
=================================
This is not related to the instalation, since moving this VPS to another machine with a different kernel makes it work perfectly.
We are almost sure this is a TS bug. Any help would be appreciated.
Thanks in Advance
How it happens?
TS starts sucessfully. People can connect to it. After some time everyone times out, the connection ont he tcp port is "accepted" but not answered. A 'ps' shows all processes are active and seem to be ok. This is not a network issue since a restart on ts makes it work again.
This is happening a lot, like once an hour.
An strace on the active processes repeats this:
=================================
select(23, [22], [], [], {0, 100000}[{fd=13, events=POLLIN}], 1, 52) = 0
gettimeofday({1149605956, 472366}, NULL) = 0
gettimeofday({1149605956, 472414}, NULL) = 0
poll() = 0 (Timeout)
=================================
This is not related to the instalation, since moving this VPS to another machine with a different kernel makes it work perfectly.
We are almost sure this is a TS bug. Any help would be appreciated.
Thanks in Advance