Bug 15161 - (apr?) socket error in cron environment
Summary: (apr?) socket error in cron environment
Status: CLOSED INVALID
Alias: None
Product: Apache httpd-2
Classification: Unclassified
Component: Core (show other bugs)
Version: 2.0-HEAD
Hardware: Other Linux
: P3 critical (vote)
Target Milestone: ---
Assignee: Apache HTTPD Bugs Mailing List
URL: http://http://
Keywords:
Depends on:
Blocks:
 
Reported: 2002-12-07 23:08 UTC by André Malo
Modified: 2004-11-16 19:05 UTC (History)
0 users



Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description André Malo 2002-12-07 23:08:47 UTC
Recently installed httpd-2.0.43 on a linux box (more or less SuSE 7.2 default) 
and it occured, that the httpd cannot be started from a cron job. It runs and 
starts normally if started manually (apachectl start) or called from the 
bootscripts (also apachectl start). apachectl restart from cronjob is also no 
problem.

If I call apachectl stop and then apache start from a cronjob, the error is:

[error] (88)Socket operation on non-socket: apr_accept: (client socket)

that fills the error log with a new entry every second or so.

Don't know, if it's a configuration error, but I don't think so. In httpd-1.3.26 
it worked fine (1.3.27 not tested). I've built also the STRIKER_2_0_44_PRE1 from 
today - same error.

Do you need some (compile) configuration data?
Comment 1 Justin Erenkrantz 2003-02-16 23:59:42 UTC
IIRC, this has something to do with some draconian ulimit's that are set in cronjobs on certain platforms.

I would suggest posting to the development list or searching the archives as I'm fairly sure this has come up before.

(Certainly, Greg Ames would be doing this on daedalus.)
Comment 2 André Malo 2003-04-10 17:25:41 UTC
Just a note to myself. Will test it, if the server's not under load ...
Comment 3 André Malo 2004-01-12 19:29:21 UTC
Had no chance to actually test it, but the server is now gone. I'm assuming, it
was an ulimit issue.