Topic: Apache 2.4.3 mod_proxy not stable in Production

[pidfile]

Merry Xmas

Have run through a bunch of threads on mod_proxy race conditions, possible causes, and potential solutions suggested by Apache core devs.

What is the status? Is there a fix for this, maybe in 2.4.7?

On 2.4.3 in production we literally have 10K errors in the past 2 months along the lines of:

[jraute]

What is the complete error line and what configuration you are using?

[Steffen]

When does it happen ?

Assuming that you use ProxyPass directive.

What about the keepalive timeout, front and back the same ? Good habbit is to make different, 1-2 seconds different (not recall front or back)

Also the default for Timeout has been reduced in 2.4 from 300 to 60. The developers said that if
there are complaints from users, the new value may be too low and we should maybe reconsider the new value.

So check and play with the settings:

TimeOut 300
KeepAlive On
MaxKeepAliveRequests 100
KeepAliveTimeOut 15

Running Windows ?
You can try:

AcceptFilter http none
AcceptFilter https none
EnableSendfile off
EnableMMAP off

[pidfile]

Thanks guys.

Linux box (CentOS 6.3)

*******************
# conf.d/foo.conf
<Location />
ProxyPass http://172.16.xx.xx:9001/ max=60 retry=10
ProxyPassReverse http://172.16.xx.xx:9001/
SetOutputFilter DEFLATE
SetEnv proxy-sendchunks 1
</Location>

# httpd.conf
KeepAlive On
KeepAliveTimeout 2500ms
EnableSendfile On
********************

Timeout shouldn't come into play, page requests typically served in < 400ms, not the 60s default.

Linux default KeepAlive timeout is 2 hours.

Again, apache dev lists do indicate that there is an as yet unresolved (as of 2.4.3 at any rate) issue with mod_proxy and keepalive race conditions.

Not sure what the underlying cause is, but getting on average at least 150 proxy errors per day _in production_ not good.

[jraute]

Although you didn't post the complete error log line i am quite sure that the problem still exists.
Reading the bug blog there seems to be only one solution which is based on deactivating keep alives and i remember such errors even with 2.4.7 in our logs too although the clients do not have any problems.

Have you tried the settings Steffen mentioned?

[pidfile]

Thanks, settings are the same as Stefan's, just shorter TTL.

Could turn off KeepAlive, but prefer performance gain, not loss.

Will dig around some more, perhaps there's something specific to my setup if clients are affected by the errors while in your case apparently they are not.

[jraute]

There's one advice in the blog to use

[pidfile]

See http://www.gossamer-threads.com/lists/apache/dev/428261 for all the gory details. From the thread:

[jraute]

Ok, no solution so far.

(Honestly spoken in our setting the "performance problem" is minimal - up to 30000 page requests per day and the performance and especially the response time is ok. But you are right, it depends on the things you want to do.)

Greets