On Fri, 25 Feb 2011 16:09:31 -0500, Greg Larkin wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 2/25/11 3:00 PM, Mark Felder wrote:
On Fri, 25 Feb 2011 13:15:29 -0600, Greg Larkin
wrote:
I haven't used watchdogd(8) (http://bit.ly/eKHUEN [1]) before, but
I
wonder
if it would help you by firing some data logging command (-e
option)
so
you have some information to go on.
That actually looks like a great solution, but I currently cannot
recompile the kernel to enable SW_WATCHDOG and take those servers
down.
We have identical hardware in a test environment that I can do that
on
and hopefully see if we can get a clone of the problematic VMs to
fail
there with the watchdogd enabled.
Thanks,
Mark
Hi Mark,
That sounds good, and please post any further information so someone
else can help you troubleshoot further.
Here are some other questions for you to ponder:
- - What appears in the Apache access & error logs just before a
freeze
occurs?
- - Are you using any unusual Apache config file settings or modules?
How
about for PHP?
- - Are all of your installed ports up to date?
- - Have you done any Apache or PHP performance tuning using
http://httpd.apache.org/docs/2.2/misc/perf-tuning.html [3] or
http://phplens.com/lens/php-book/optimizing-debugging-php.php [4] as
starting points?
- - Have you enabled a PHP log file to record errors, warnings, etc.
generated by your application code?
Hope that helps,
Greg
- --
Greg Larkin
http://www.FreeBSD.org/ [5] - The Power To Serve
http://www.sourcehosting.net/ [6] - Ready. Set. Code.
http://twitter.com/sourcehosting/ [7] - Follow me, follow you
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ [8]
iEYEARECAAYFAk1oGosACgkQ0sRouByUApAQOgCfW47AXRKSCuj2ftZJSQubVEPw
w54Ani/7k+SNvR4JllG5wJXQSmBE90Be
=1PvS
-----END PGP SIGNATURE-----
_______________________________________________
[email protected] [9] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions [10]
To unsubscribe, send any mail to
"[email protected] [11]"
I manage 8 FreeBSD virtual servers at work running 7.0, 7.1, and 7.2 in
VMware ESX 4.0, they are running Apache2 with PHP, Bind, and Squid, any
of the services they are running is all installed from source and kept
up to date, as we are required to pass quarterly PCI scans. So it
doesn't speak much of programs installed from ports, but these systems
stay very stable, the only issue I have is that after about 8 to 9
months of system uptime (services are restarted more frequently for
updates) sometimes the network will stop responding until the virtual
machine is shutdown and powered back on. (I can't say that any of the
windows servers on the cluster ever reach that length of uptime to
compare with this) We have 6 physical servers running with full
automated vmotion there is about 80 windows servers runnign on the same
systems. Backed with an ISCSI SAN and 10G Ethernet adapters. However
our internet applications that utilize the FreeBSD servers are
relatively low usage, squid is mainly used as outbound proxy for around
600 Computer users which is the highest used resource on these servers,
primarily between 5am and 7pm, so usage levels may not be near where you
are running at.
I am running vmware tools installed from a VMware workstation 6.5 tools
image, I have switched a couple of the systems over to the vmxnet
drivers instead of running then as e1000 adapters, but haven't been
running them long enough to see if that resolves are problem. When it
does occur the system is completely responsive when connecting to the
console though, and can be cleanly shutdown.
I haven't noticed any issues with VMwawre performance monitoring
reporting the systems CPU as hammered when the CPU on FreeBSD shows low,
but these systems are much more heavily memory used followed by disk I/O
CPU sits idle at most times. I should also mention that they are setup
with 2 vCPUs, perhaps if you are running a signal vCPU something is
hanging one vCPU and mine setup is recovering OK because the
applications complete using the other until the issue resolves itself.
I haven't spent much time looking at historical usage graphs on these
since I don't get any complaints about performance.
---
Thanks,
Dean E. Weimer
[email protected]
http://www.dweimer.net/
_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[email protected]"
