Description of Problem: Error logs are accumulating in the /var/log/mailman directory. They have names like this: -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.2.4.1.2 -rw-rw-r-- 1 mailman mailman 0 Sep 29 05:01 error.1.1.1.1.1.1.3.2.4.1.2.1 -rw-rw-r-- 1 mailman mailman 0 Sep 15 04:05 error.1.1.1.1.1.1.3.2.4.1.3 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.2.4.2 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.2.4.2.1 -rw-rw-r-- 1 mailman mailman 0 Sep 29 05:01 error.1.1.1.1.1.1.3.2.4.2.1.1 -rw-rw-r-- 1 mailman mailman 0 Sep 22 04:16 error.1.1.1.1.1.1.3.2.4.2.2 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.2.4.3 -rw-rw-r-- 1 mailman mailman 0 Sep 29 05:01 error.1.1.1.1.1.1.3.2.4.3.1 -rw-rw-r-- 1 mailman mailman 0 Sep 8 04:03 error.1.1.1.1.1.1.3.2.4.4 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.3 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.3.1 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.3.1.1 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.3.1.1.1 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.3.1.1.1.1 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.3.1.1.1.1.1 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.3.1.1.1.1.1.1 -rw-rw-r-- 1 mailman mailman 0 Sep 29 05:01 error.1.1.1.1.1.1.3.3.1.1.1.1.1.1.1 -rw-rw-r-- 1 mailman mailman 0 Sep 22 04:16 error.1.1.1.1.1.1.3.3.1.1.1.1.1.2 -rw-rw-r-- 1 mailman mailman 0 Oct 6 07:37 error.1.1.1.1.1.1.3.3.1.1.1.1.2 -rw-rw-r-- 1 mailman mailman 0 Sep 29 05:01 error.1.1.1.1.1.1.3.3.1.1.1.1.2.1 -rw-rw-r-- 1 mailman mailman 0 Sep 15 04:05 error.1.1.1.1.1.1.3.3.1.1.1.1.3 These files are present only on some servers, but not on others. Servers running RedHat 7.3 seem to be the ones that contain these files. OTHER SYMPTOMS: Sometimes, logrotate seems to run and take up a tremendous amount of CPU and disk (albeit for only about 2 to 10 minutes. I am assuming that this is caused by it dealing with all these mailman error files.) Version-Release number of selected component (if applicable): logrotate-3.6.4.1 mailman-2.0.9-1, mailman-2.0.13-1 Happens with the same version of logrotate and several versions of mailman. How Reproducible: Seems to consistently happen under RedHat 7.3 (all servers I've looked at.) It definately does NOT have something to do with root: afreed in /etc/aliases, because this entry is NOT present on my laptop, which is experiencing the same problem. Steps to Reproduce: 1. Problem appears to be present on Redhat 7.3 servers that have the aforementioned versions of mailman and logrotate. 2. Cannot remove the logs through conventional means as this results: [root@montpellier root]# cd /var/log/mailman [root@montpellier mailman]# rm -Rf * bash: /bin/rm: Argument list too long [root@montpellier mailman]# ls -al | wc -l 31675 31675 is not a huge number of files. Unless RM is hardwired to not remove files in a directory with the s flag set for Groups, then this might well be a bug with RM. (I CAN remove the files one-by-one or a few at a time.) 3. Actual Results: error files accumulate, eventually to the point where all available inodes are used up and no more data can be stored in /var/log/, which, in turn, can wedge the server, cause scripts to fail and otherwise wreak mayhem and contumily. Expected Results: The files should be cleaned up and removed by normal logrotate operation. It seems to be unique to mailman errors, though. I have seen no other logsfiles accumulating in any other directories. Additional Information: Available upon request.
Created attachment 79229 [details] Directory listing of "stuffed" /var/log/mailman
This is a bug with the confiog files in redhat - you need to edit /etc/logrotate.d/mailman from "/var/log/mailman/*" to explicitly name the files you want to rotate eg "/var/log/mailman/error" I had 350,000 logfiles in mailman this morning - I've been killing off logrotate lately as it's been hogging all the CPU , now I know why. Check your mgetty setup if your using it too - it has a wildcard in it as well.
I've diagnosed this problem on at least 2 RH 7.3 systems (and I assume it occurs on all of them). The system eventually runs out of inodes on /var and chews up huge amounts of CPU when doing log rotation. Processes such as dhcpd and syslog that both periodically write to /var fail since they can't create a file. Inodes increase logarithmically, which explains why my RH 7.3 have/will experience this problem at roughly the same time.
Here's a command line for removing the files. I'm sure there is more than one way to do it. find /var/log/mailman -name error\* -exec /bin/rm {} \; This works on the bash command line - your shell may require different quote/escape chars.
*** This bug has been marked as a duplicate of 68917 ***