Mailman and Internal Server Errors

[sparek-3]

I am having some problems with the Mailman software installed on the server. A few clients have written in about Internal Server errors when they try to log into the the list's administrative section. I have verfied that this is the case. However this is not server-wide, some lists will work, other's don't. When reviewing the Apache error log, all I see is:

Premature end of script headers: /usr/local/cpanel/3rdparty/mailman/cgi-bin/admin

I am able to fix the issue by running /scripts/reinstallmailman. However this just does not appear to be a permanent fix. While it will fix one list, but either breaks other lists or does not fix them all, as clients will continue to report in with the same Internal Server Error for different lists. This is becoming quite a hassle to reinstall Mailman everytime a client reports a broken mailing list and I suspect that the issue is somewhere deeper.

I did notice some other threads concerning this issue, but they were all dated from last year or later, so I'm not sure if my current issue and the issues discussed in those threads are related. I also just do not see a definitive answer in any of those threads. One of the larger threads centers around if you moved the mailman stuff out of the /usr partition, which I have not done.

I have also tried:

/scripts/fixmailman
/scripts/fixmailmanwithsuexec
/scripts/fixoldlistswithsuexec

But none seem to fix the issue. I currently have one client who is experiencing this problem. Rather than run /scripts/reinstallmailman again, I thought I would see if anyone here has any suggestions on a more permanent fix.

It should also be noted that the logs in /usr/local/cpanel/3rdparty/mailman/logs do not say anything when the Internal Server Error is encountered. The permissions for the /usr/local/cpanel/3rdparty/mailman/cgi-bin/admin file is set to 2755. I don't think its an issue with the admin binary, because other lists are working on the same server. I am running 10.8.1-RELEASE_113. If there is any other information that you need let me know, I've tried to think of everything that I can. Maybe its just a setting that I am overlooking and somebody can point it out.

Thanks

[lloyd_tennison]

A couple of quick items.

1. Everytime you reinstall mailman - it deletes all your logs. Not a good thing.

2. My mailman files in cgi-bin are all -rwxr-sr-x. See thread: http://mail.python.org/pipermail/mai...er/046593.html

3. Try running check_perms and check_db and see what happens.

[sparek-3]

Thanks for the reply. I actually just did a mailman reinstall again, because I wasn't sure if there were going to be any suggestions, and I was intentionally leaving this client's list inaccessible, while I tried to determine what the underlying issue.

I did not know that about the mailman logs. I don't figure it is all that big of a deal to lose the logs, but still I did not know this, so thanks for the info and I'll remember that.

All of the files in the mailman/cgi-bin are as you stated (2755) except for the create binary, its got permissions of 000. Not sure why that is the case, but I did compare this to another one of our servers and they match up, the create binary is 000 on that server as well. The create cgi is not listed in the error log and I really don't think it has anything to do with the permissions on these files because some (most) lists work, its just every now and then a client will write in with this problem and lately we have been receiving these much more often.

I did run check_perms and check_db on the list this morning. check_perms reported that everything was fine and check_db did not find any problems either. I was still not able to log in after running these.

When I was reinstalling mailman, I did see a mention of a lock file with the list in question mentioned. I don't know what mailman does in terms of lock files. But I believe this may be something to check out if I encounter this issue again. I'm not sure where the lock file was located, and like I said I've reinstalled mailman so this list is now working, so I don't have a non-working list to try this on. A lock file issue does seem to fit, as it would explain why every other list is working fine except for one particular list. However, I just really don't know when and where mailman does any lock files.

If I encounter this problem again, I'll look for a lock file and see if removing the lock file helps. I'll update this thread if I find anything useful.

For anyone else that may be encountering this problem, you might try looking to see if a lock file is present. I suspect the lock files would be in /usr/local/cpanel/3rdparty/mailman/locks and should resemble something like listname_domain.com.lock or listname_domain.com.lock*. If a lock file is not in this directory try changing into /usr/local/cpanel/3rdparty/mailman and running:

find -name listname_domain.com.lock* -print

I would be interested to know if a lock file is indeed causing these issues.

[lloyd_tennison]

Actually the subscribe log is very important. Without it you cannot say for certain how a person got on the list and that is the only way to satisfy spam complaints.

[sparek-3]

Just to follow up on this issue. This issue was caused by a lock file in the /usr/local/cpanel/3rdparty/mailman/locks directory. The lock file uses the syntax listname_domain.com.lock. If you remove that lock file, then you can access the administrative portion of that list. I'm not sure what causes the lock file to be created or why it is not removed.