LSAPI Sockets Closing Periodically

[Ghan_04]

I am having an issue where I have a few sites setup using PHP with the LSAPI external app. Each one has its own external app config so that I can use suEXEC and have each site run its own PHP processes. However, since switching over to this config, I am finding occasional 503 service unavailable errors with this showing up in the error log:

2012-02-19 04:10:41.188 [INFO] [66.249.71.161:42504-0#wayward] connection to [/tmp/lshttpd/wayward_lsphp.sock] on request #0, confirmed, 0, associated process: 5264, running: 1, error: No such file or directory!
2012-02-19 04:10:41.188 [INFO] [uds://tmp/lshttpd/wayward_lsphp.sock] Connection refused, restart!
2012-02-19 04:10:41.189 [INFO] [66.249.71.161:42504-0#wayward] connection to [/tmp/lshttpd/wayward_lsphp.sock] on request #0, confirmed, 0, associated process: 5264, running: 1, error: No such file or directory!
2012-02-19 04:10:41.189 [INFO] [uds://tmp/lshttpd/wayward_lsphp.sock] Connection refused, restart!
2012-02-19 04:10:41.189 [INFO] [66.249.71.161:42504-0#wayward] connection to [/tmp/lshttpd/wayward_lsphp.sock] on request #0, confirmed, 0, associated process: 5264, running: 1, error: No such file or directory!
2012-02-19 04:10:41.189 [INFO] [uds://tmp/lshttpd/wayward_lsphp.sock] Connection refused, restart!
2012-02-19 04:10:41.189 [NOTICE] [66.249.71.161:42504-0#wayward] Max retries has been reached, 503!
2012-02-19 04:10:41.189 [NOTICE] [66.249.71.161:42504-0#wayward] oops! 503 Service Unavailable
2012-02-19 04:10:41.189 [NOTICE] [66.249.71.161:42504-0#wayward] Content len: 0, Request line: 'GET /find-new/494107/threads?_debug=1 HTTP/1.1'
2012-02-19 04:10:41.189 [NOTICE] [66.249.71.161:42504-0#wayward] Redirect: #1, URL: /index.php

The lsphp.sock files appear to be getting closed and vanishing periodically so that they have to be restarted. I am also seeing lots of these statements in the log:

2012-02-19 04:10:12.000 [INFO] [CLEANUP] Clean up child process with pid: 5229
2012-02-19 04:10:12.000 [INFO] [CLEANUP] Clean up child process with pid: 5231
2012-02-19 04:10:12.006 [INFO] Remove pid: 5229, exitcode: 0
2012-02-19 04:10:12.010 [INFO] Remove pid: 5231, exitcode: 0
2012-02-19 04:10:22.000 [INFO] [CLEANUP] Clean up child process with pid: 5260
2012-02-19 04:10:22.008 [INFO] Remove pid: 5260, exitcode: 0

I have the external app set to auto start and run on startup with persistent connections turned on.
Any ideas how I can resolve these errors?

Thanks!

 

[Ghan_04]

Ok so I think I found the setting that "fixes" this.
The external app's "Max Idle Time" should be set to -1 so that none of the sockets get killed to free up resources.
However, this results in a lot of child lsphp5 processes stacking up taking up RAM, which I would like to avoid if possible.

It sounds like this might be a bug - the fact that the socket gets killed to free up resources should not result in a 503 next time someone requests a page from that virtual host. Is there another configuration setup that I'm missing that fixes this?

Thanks.

 

[NiteWave]

what's other settings in lsphp ?
Max Connections
Connection Keepalive Timeout
Instances

for
Run On Start Up,
you can set it to No to see if any difference.

 

[Ghan_04]

Max Connections = 5
Connection Keepalive Timeout = Blank
Instances = 1

Changing Run on Startup to No with my previous value of Max Idle Time (10) did not seem to have any effect - I'm seeing the same messages in the error log.

 

[NiteWave]

forgot to ask

Environment

setting in Ext App.

 

[Ghan_04]

Right now I've got it set to this:

PHP_LSAPI_MAX_REQUESTS=10
PHP_LSAPI_CHILDREN=5

I had the max requests set much higher earlier - it doesn't seem to make much difference.