Project

General

Profile

Actions

Bug #6632

closed

alternative to using -X while restarting apache on the rpms

Added by Tamilarasi muthamizhan over 10 years ago. Updated over 10 years ago.

Status:
Resolved
Priority:
Urgent
Assignee:
Sandon Van Ness
Category:
-
% Done:

0%

Source:
Q/A
Tags:
Backport:
Regression:
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Crash signature (v1):
Crash signature (v2):

Description

In the teuthology rgw task, -X is being used while restarting apache , so the server is not detached from the console for the subsequent runs, which is important for multi region tests.

but this affects the rgw test runs on rpms [rhel/centos/f18]. getting rid of -X works with rpms.

we are looking at finding an alternative to this one, one of which was to replace -X with -DNO_DETACH which still did not work.

another option: fork another apache2 process and do a killall towards the end of test.

any other thoughts?

Actions #1

Updated by Sandon Van Ness over 10 years ago

Tamil asked me to look into this.

It took some digging but I think I figured this out.

Testing on ubuntu it appears doing multiple connections when running with -X nothing is blocked (all in a single process). I assume it times out on rhel, centos, etc.. because in my testing with the same config file when I open a second connection it is blocked until the first one is stopped. I think I traced this to worker.c not being compiled in httpd:

ubuntu:

ubuntu@plana05:~$ apache2 -l
Compiled in modules:
core.c
mod_log_config.c
mod_logio.c
mod_version.c
worker.c
http_core.c
mod_so.c

Centos:

[root@vpm010 ~]# httpd -l
Compiled in modules:
core.c
prefork.c
http_core.c
mod_so.c

So anyway I found out centos has a seperate httpd (httpd.worker) which has worker.c copmiled in instead of the default (prefork):

[root@vpm010 ~]# httpd.worker -l
Compiled in modules:
core.c
worker.c
http_core.c
mod_so.c

When testing running using httpd.worker (instead of http) I am not getting blocked requests like before and I am able to run with -X so I think simply changing teuthology to use httpd.worker instead of just 'httpd' might get around this issue. Let me know.

Actions #2

Updated by Josh Durgin over 10 years ago

That sounds like a good fix to me.

Actions #3

Updated by Sage Weil over 10 years ago

  • Priority changed from High to Urgent
Actions #4

Updated by Sage Weil over 10 years ago

  • Assignee set to Sandon Van Ness
Actions #5

Updated by Sage Weil over 10 years ago

  • Status changed from New to In Progress
Actions #6

Updated by Sage Weil over 10 years ago

  • Status changed from In Progress to Resolved
Actions

Also available in: Atom PDF