Bug 17188

Summary:	Syslog takes 100% of the CPU when binded to an non-existant socket.
Product:	[Retired] Red Hat Linux	Reporter:	Guillaume Filion <gfk>
Component:	sysklogd	Assignee:	Bill Nottingham <notting>
Status:	CLOSED DUPLICATE	QA Contact:
Severity:	high	Docs Contact:
Priority:	medium
Version:	6.1	CC:	rvokal
Target Milestone:	---
Target Release:	---
Hardware:	i386
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2000-09-02 14:54:28 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Guillaume Filion 2000-09-02 14:54:24 UTC

Bastille-Linux changed my /etc/rc.d/init.d/syslog file and added this 
command to start syslogd:
daemon syslogd -m 0 -a /home/dns/dev/log

However, as I was not using Bind, I deleted /home/dns. This caused 
syslogd to go berzerk on next restart:
When syslogd starts, I get this in /var/log/messages:

Aug 24 09:53:19 sanitized syslogd 1.3-3: restart.
Aug 24 09:53:19 sanitized syslogd: select: Bad file descriptor
Aug 24 09:53:50 sanitized last message repeated 416955 times
Aug 24 09:54:51 sanitized last message repeated 860294 times
Aug 24 09:55:52 sanitized last message repeated 869058 times
Aug 24 09:56:53 sanitized last message repeated 870385 times
Aug 24 09:57:54 sanitized last message repeated 869721 times
Aug 24 09:58:55 sanitized last message repeated 869977 times
Aug 24 09:59:56 sanitized last message repeated 869792 times
Aug 24 10:00:57 sanitized last message repeated 862309 times
Aug 24 10:01:58 sanitized last message repeated 851399 times
Aug 24 10:02:59 sanitized last message repeated 862334 times
Aug 24 10:04:00 sanitized last message repeated 859975 times
Aug 24 10:05:01 sanitized last message repeated 859789 times
Aug 24 10:06:02 sanitized last message repeated 863366 times
Aug 24 10:07:03 sanitized last message repeated 868746 times
Aug 24 10:08:04 sanitized last message repeated 868313 times
Aug 24 10:09:05 sanitized last message repeated 868068 times
Aug 24 10:10:06 sanitized last message repeated 867668 times
Aug 24 10:11:07 sanitized last message repeated 870930 times
Aug 24 10:12:08 sanitized last message repeated 870589 times
Aug 24 10:13:09 sanitized last message repeated 870940 times
Aug 24 10:14:10 sanitized last message repeated 870386 times
Aug 24 10:15:11 sanitized last message repeated 871397 times
Aug 24 10:16:12 sanitized last message repeated 877305 times
Aug 24 10:17:13 sanitized last message repeated 876981 times
Aug 24 10:18:14 sanitized last message repeated 877334 times

and my CPU load peeks up at about 10 until syslogd or the kernel 
crashes. This thing would DOS myself.

Removing the "-a /home/dns/dev/log" options solved the problem.

This looks like an infinite loop problem when syslogd is binded to an 
non-existant socket.

Comment 1 Bill Nottingham 2000-10-02 22:16:15 UTC


*** This bug has been marked as a duplicate of 10363 ***