Hi,
We're experiencing problems with rogue processes when using APC php caching in combination with the dt Apache agent (dynaTrace version 6.0.0.7, Apache 2.2)
When we use either the Apache agent or APC, everything works fine, but when we combine them both and we try to do a 'httpd graceful' (like exectuted by our 'rotate log' cron job), the old httpd child process that has an open cache file in /tmp fails to exit.
We can clearly see that the status of the open file is 'deleted', but because the process never exits, the disk space is never freed.
After a while, we have a lot of those httpd rogue processes and when left unattended, /tmp gets completely filled and our webserver fails.
Killing a rogue process immediately frees its disk space and executing a full 'httpd restart' kills all rogue processes, but most of the time we want to be able to use
'httpd graceful'.
We've tried using a php agent instead (to be 100% clear: without Apache agent) and that works fine (so no httpd rogue processes when doing a 'httpd graceful'), but not all of the requests we're interested in are php requests and we still want to be able to use UEM script injection, so that's no real solution in our case.
Is there anyone who experienced something similar with Apache agents and maybe has found a better solution than the inconvenient work-arounds I've already mentioned (keep manually killing processes, stop using 'graceful', try to use a pure php agent set-up)?
Thanks in advance for any feedback,
Wouter
Answer by Christian G. ·
Hi Wouter!
There were some reports of hanging Apache worker processes in 6.0 and we made fixpacks for those hangs. Would you mind contacting our support team? They would help you to find a fixpack for you or to gather all the information we need to analyze that problem.
Regards,
Christian.
JANUARY 15, 3:00 PM GMT / 10:00 AM ET