Hi,
we do have the following problem:
the HTTP Proxy will stop working after some hours (sometimes it's a few days). The result is this:
The proxy comsumes > 99% CPU time (top), the load of the machine goes up to 53 !! (uptime) and new TCP Sessions to port 8080 will see a RESET (sniffed at the firewall).
We disabled almost all options that could have caused the problem (except AV scanning). Same problem.
After a restart of the proxy (/var/mdw/scripts/httpproxy restart), the proxy runs fine and with good performance for a couple of hours/days. Then the same game starts again.
If the proxy hangs, there are no more log entries in /var/log/http.log, except these:
2009:04:29-15:04:42 astaro-2 httpproxy[5377]: [ (nil)] sc_check_servers (scr_scanner.c:722) server 'cffs01.astaro.com' access time: 23ms
2009:04:29-15:04:42 astaro-2 httpproxy[5377]: [ (nil)] sc_check_servers (scr_scanner.c:722) server 'cffs02.astaro.com' access time: 32ms
etc.
We did some log analysis to check if the problem is caused by some clients or too much requests. However, there are very few requests per second (5-10), when the problem occurs and just a few clients active (1-5).
We are not able to identify any hardware problem as the problem source and suspect it could be a problem with the proxy software itself (bug). But then we wonder, why there are not more reports about such a problem.
Any idea how to troubleshoot this issue or what it could cause? Remember, the proxy does not accept new connections when it goes mad, but it runs at > 99% cpu time and causes a load of 53 !! on the machine. WHAT the hell is it doing???
Thanks
Al Gorithm
This thread was automatically locked due to age.
