All,
We're running some performance tests against some web
based applications using Apache 2.2 on Solaris 10 (u3), using IPFilter
4.1.24 and PFIL 2.1.13. During the course of multiple tests, we've seen
active HTTP connections, as well as SSH sessions to the machine 'lock
up', and fail to respond. You can establish new connections, and those
will work for a short period of time, and then they will lock up as
well. SSH connections are eventually terminated do to loss of
communication. In an effort to determine if this was a IPFilter or some
other OS problem, we performed an 'ifconfig <interface> modremove
[EMAIL PROTECTED]', and the issue did not occur during subsequent tests on a
V100
(dmfe interfaces).
We switched our testing over to the Sun T1000's (bge interfaces) that we
will be using for production deployments, and saw the same situation
there. However, I was able to issue an ifconfig modremove/modload pair
in short order on this machine, and was able to recover the connections
prior to SSH tearing them down (the HTTP requests were subsequently
issued on new connections and recovered on their own).
In neither situation did we see any messages to the console,
/var/adm/messages, or /var/log/syslog that would indicate other problems
on the machines.
Has anyone seen this behavior before? Is there a configuration
parameter in the kernel I need to look at, or could I have introduced
something during the compile process? Any other thoughts or feedback?
Thanks in advance for the help.
Scott Hopkins
