apache?
Kenny Lussier
klussier at gmail.com
Sat Apr 4 13:15:43 EDT 2009
Hi All,
I am seeing an interesting problem with what I think is apache, but could be
tcp related as well. Our firewall blocks all connections to 443 except for
specific hosts. When someone connects to apache on 443, we proxy the
connection to a load balancer, which sprays the connections across several
tomcat servers.
A few days ago, we upgraded the front-end apache servers from RHEL3 running
httpd-2.0.47 to RHEL5, httpd-2.2.3-22. Since then, we have been experiencing
strange intermittant outages. We see connections to our firewall on port
443, and we see SOME traffic on the apache servers, but most people trying
to connect are timing out. This lasts for about 5 minutes. There are no
errors logged in messages or in apache's error_log. However, seconds AFTER
the problem clears up, I see:
[Fri Apr 03 21:37:42 2009] [info] server seems busy, (you may need to
increase StartServers, or Min/MaxSpareServers), spawning 8 children, there
are 0 idle, and 107 total children
[Fri Apr 03 21:37:43 2009] [info] server seems busy, (you may need to
increase StartServers, or Min/MaxSpareServers), spawning 16 children, there
are 3 idle, and 115 total children
[Fri Apr 03 21:37:44 2009] [info] server seems busy, (you may need to
increase StartServers, or Min/MaxSpareServers), spawning 32 children, there
are 11 idle, and 131 total children
[Fri Apr 03 21:37:45 2009] [info] server seems busy, (you may need to
increase StartServers, or Min/MaxSpareServers), spawning 32 children, there
are 42 idle, and 163 total children
[Fri Apr 03 21:57:35 2009] [info] server seems busy, (you may need to
increase StartServers, or Min/MaxSpareServers), spawning 8 children, there
are 46 idle, and 128 total children
[Fri Apr 03 21:57:43 2009] [info] server seems busy, (you may need to
increase StartServers, or Min/MaxSpareServers), spawning 8 children, there
are 48 idle, and 146 total children
There is no change in traffic volume, there have been no network changes,
either. One thing that I noticed is that in RHEL3, tcp_syncookies is set to
0 (off) by default, and on RHEL5 it is set to 1 (on). Could syncookies be
causing this? Has anyone else ever experienced anything like this?
TIA,
Kenny
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.gnhlug.org/mailman/private/gnhlug-discuss/attachments/20090404/3bfaba52/attachment.html
More information about the gnhlug-discuss
mailing list