-
Hi, I'm having an issue on an environment with a large number of virtual hosts, up to 600 in some cases. A number of these vhosts use mod_auth_openidc with Entra ID and we have a separate .conf include for each:
The problem we have is that with ~200 of these .conf files included, Apache is unable to start. top shows httpd running for the first few seconds after restarting with gradually increasing virtual memory use until it gets to ~64GB, it then fails with out of memory errors. With only a few of these .conf includes in place, everything works perfectly and memory use is KB. Would you expect this kind of behaviour with a large number of vhosts that include openidc config or is there something we're doing in the config above that might be causing this? Thanks! |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 3 replies
-
we recently bumped into this in our CI/CD environment with similar numbers; it turned out to be the Linux OOM Killer that kills the apache processes because of memory consumption ( we minimized the it also appears that the |
Beta Was this translation helpful? Give feedback.
-
after some more testing it appears that the initialization issue is actually non-existent; so the advice - if you have a large number of vhosts - is: |
Beta Was this translation helpful? Give feedback.
after some more testing it appears that the initialization issue is actually non-existent; so the advice - if you have a large number of vhosts - is:
a) add more memory to the server running Apache
or, if a) is not possible for. you:
b) tune the cache with
OIDCCacheShmMax
andOIDCCacheShmEntrySizeMax
and possibly also reduce the number of Apache processes ona single hostor, if b) does somehow not work for you:
c) use a cache backend other than
shm
, e.g.redis