apache bad file descriptor

deesto · 10-23-2007, 10:09 AM

I'm seeing this error message each time my Apache server is restarted with apachectl graceful:

Code:

[Tue Oct 23 10:28:01 2007] [notice] Graceful restart requested, doing restart
[Tue Oct 23 10:28:01 2007] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Tue Oct 23 10:28:01 2007] [notice] Apache/2.2.6 (Unix) mod_ssl/2.2.6  configured -- resuming normal operations

The error isn't fatal: the server does start, as seen in the last line. Nothing in messages except for dropped iptables input.

System: RHEL4u5 Linux, 2 PIII 1 GHz CPUs, 1 GB RAM.

Apache details (built from latest available tar archive):

Code:

]# apachectl -V
Server version: Apache/2.2.6 (Unix)
Server built:   Oct 22 2007 10:22:43
Server's Module Magic Number: 20051115:5
Server loaded:  APR 1.2.11, APR-Util 1.2.10
Compiled using: APR 1.2.11, APR-Util 1.2.10
Architecture:   32-bit
Server MPM:     Prefork
  threaded:     no
    forked:     yes (variable process count)
Server compiled with....
 -D APACHE_MPM_DIR="server/mpm/prefork"
 -D APR_HAS_SENDFILE
 -D APR_HAS_MMAP
 -D APR_HAVE_IPV6 (IPv4-mapped addresses enabled)
 -D APR_USE_SYSVSEM_SERIALIZE
 -D APR_USE_PTHREAD_SERIALIZE
 -D SINGLE_LISTEN_UNSERIALIZED_ACCEPT
 -D APR_HAS_OTHER_CHILD
 -D AP_HAVE_RELIABLE_PIPED_LOGS
 -D DYNAMIC_MODULE_LIMIT=128
 -D HTTPD_ROOT="/etc/httpd"
 -D SUEXEC_BIN="/etc/httpd/bin/suexec"
 -D DEFAULT_PIDLOG="logs/httpd.pid"
 -D DEFAULT_SCOREBOARD="logs/apache_runtime_status"
 -D DEFAULT_LOCKFILE="logs/accept.lock"
 -D DEFAULT_ERRORLOG="logs/error_log"
 -D AP_TYPES_CONFIG_FILE="conf/mime.types"
 -D SERVER_CONFIG_FILE="conf/httpd.conf"

trickykid · 10-23-2007, 11:15 AM

Do you get the same error if you do a stop and start?

trickykid · 10-23-2007, 11:20 AM

You might be hitting this bug:

http://issues.apache.org/bugzilla/show_bug.cgi?id=42829

I'd suggest just doing a full stop and start to get around the error until fixed. I never do graceful restarts with apache, never seems to do the job accurately.

deesto · 10-23-2007, 11:23 AM

Quote:

Originally Posted by trickykid

Do you get the same error if you do a stop and start?

Hmmmm ... nope:

Code:

[Tue Oct 23 12:20:13 2007] [notice] caught SIGTERM, shutting down
[Tue Oct 23 12:20:15 2007] [notice] Apache/2.2.6 (Unix) mod_ssl/2.2.6 OCOL configured -- resuming normal operations

Interesting! Not good, but interesting.

Quote:

Originally Posted by trickykid

I'd suggest just doing a full stop and start to get around the error until fixed. I never do graceful restarts with apache, never seems to do the job accurately.

I think you're right. That bug seems familiar. I've been using "graceful" for some time, because I was told it's the "right" way to restart. But as you said, there seem to be lots of reasons for doing a hard stop/start instead.

The bug page you linked included a patch file. Do you think it's worth trying such a patch? Does Apache follow up on such bugs quickly?

trickykid · 10-23-2007, 01:15 PM

Apache seems to stay updated but I think they try to roll a lot of changes and fixes in at once, so they're not rolling out new fixes every other day. That would put a strain on admins.

I'd say wait for a release, if it's not security related and there's a work around, that's how I deal with such things.