[Mailman-Users] web interface bug after restore mailman

Cen Gao wrote"

Look in mm_cfg.py for a definition of LOG_DIR. If there isn't one, then
look in Defaults.py ( there it should be
LOG_DIR = os.path.join(VAR_PREFIX, 'logs')
), and if so, then look in Defaults.py for the definition of
VAR_PREFIX. This should tell you where the logs are.

If you still can't find them, or if the problem is the error log isn't
being written (are the smtp and post logs in the same directory being
written?), you can edit $mailman/scripts/driver and find line 31 or so
which says

STEALTH_MODE = 0

and change it to

STEALTH_MODE = 1

which will display the error information in your browser. Without this
information, there's not much more we can do.

Hi Mark:
thanks a lot for your suggestion.
the error log is as follows


Bug in Mailman version 2.1.9

We're sorry, we hit a bug!

If you would like to help us identify the problem, please email a copy
of this page to the webmaster for this site with a description of what
happened. Thanks!

Traceback:


Traceback (most recent call last):
File "/usr/local/bin/mailman//scripts/driver", line 82, in run_main
immediate=1)
File "/usr/local/bin/mailman//Mailman/Logging/StampedLogger.py",
line 52, in __init__
Logger.__init__(self, category, nofail, immediate)
File "/usr/local/bin/mailman//Mailman/Logging/Logger.py", line 50, in __init__
self.__get_f()
File "/usr/local/bin/mailman//Mailman/Logging/Logger.py", line 68, in __get_f
1)
File "/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/codecs.py",
line 566, in open
file = __builtin__.open(filename, mode, buffering)
IOError: [Errno 13] Permission denied: '/usr/local/bin/mailman/logs/error'
Python information:

Variable Value
sys.version 2.3.5 (#1, Dec 25 2005, 07:24:19) [GCC 4.0.1 (Apple
Computer, Inc. build 5250)]
sys.executable /usr/bin/python
sys.prefix /System/Library/Frameworks/Python.framework/Versions/2.3
sys.exec_prefix /System/Library/Frameworks/Python.framework/Versions/2.3
sys.path /System/Library/Frameworks/Python.framework/Versions/2.3
sys.platform darwin



FYI, the log directory and even all the mailman directory is owned by
mailman:mailman

And in /logs/error , it says


Aug 01 00:02:04 2007 (432) Uncaught runner exception: multipart message with
no defined boundary
Aug 01 00:02:04 2007 (432) Traceback (most recent call last):
File "/usr/local/bin/mailman//Mailman/Queue/Runner.py", line 100, in _oneloop
msg, msgdata = self._switchboard.dequeue(filebase)
File "/usr/local/bin/mailman//Mailman/Queue/Switchboard.py", line 164, in dequ
eue
msg = email.message_from_string(msg, Message.Message)
File "/usr/local/bin/mailman/pythonlib/email/__init__.py", line 51, in message
_from_string
return Parser(_class, strict=strict).parsestr(s)
File "/usr/local/bin/mailman/pythonlib/email/Parser.py", line 75, in parsestr
return self.parse(StringIO(text), headersonly=headersonly)
File "/usr/local/bin/mailman/pythonlib/email/Parser.py", line 64, in parse
self._parsebody(root, fp, firstbodyline)
File "/usr/local/bin/mailman/pythonlib/email/Parser.py", line 246, in _parsebo
dy
raise Errors.BoundaryError(
BoundaryError: multipart message with no defined boundary
Aug 01 00:02:04 2007 (432) Ignoring unparseable message: 1181090435.8718021+ee06
64d951288e0b7e6fb4b4e2b0efdc18362709

and this error log has only been written twice since I restored this
directory, although I've accessed the web page tons of times.


thanks again for your help


Cen

The web server cannot write the error log. Are all the wrappers in
$mailman/cgi-bin/ group mailman and do they all have the SETGID bit
set?

E.g., -rwxr-sr-x xxxx mailman


If this is OK, I seem to recall some Apple issue mentioned on this list
a couple of years ago. Search the archives.
<http://www.python.org/cgi-bin/faqw-mm.py?req=show&file=faq01.018.htp>


Was there anything else on the page before this about the initial
problem that it was trying to write to the error log?



This error is from an incoming post that is unparseable - almost
certainly spam. The message was discarded.


The web server can't write the error log.

Hi Mark:
thanks again for your help
Yes, all the wrappers in cgi-bin have -rwxr-sr-x mailman:mailman.
Does this bug come out just because of the webserver cannot write error logs?
It seems that everything in the logs/ directory has been written
correctly. (smtp,post,etc.) Ever the error log, if I mistakenly type
in bin/mailmanctl without any argument,this action was immediately
written this in the error log

no, I paste every line in the trackback & python informtaion sections here.

And the weird thing is mailman works prefectly well even yesterday
morning on exactly the same machine. Then I backup it, reinstall the
OS, copy it back, Set the httpd and cron. Now the mailman works but
just the web interface down.I carefully checked every singe directory
to make sure the file owner and privilege are exactly the same as
before.
I really have no clue here.

Yeah, checked the pervious error log, it seems those same errors have
been written all the time even when mailman works well. So I don't
think that will cause a problem either.

Hi Mark:

Thanks for your help. Never mind, now it works
I simply remove the log file and let mailman regenerate it, just as
you advised in a previous email on this list.
Now the owner is www instead of mailman, and the web interface is back online.
I thought it might be the group +w thing that cause www user cannot
write to error log.
Sorry that I should check it more carefully.


Thanks

Cen

This is a workaround. It is not the proper fix for the underlying
problem. The log files should not need to be owned by the web server
user because the wrappers are group mailman and SETGID, y the time the
web server wants to write a log, et should be executing with effective
GID 'mailman' and the logs should be group mailman and group writable.

Sorry, part of that was a bit garbled. It should say "... at the time
the web server wants to write a log, it should be executing ..."

thanks mark.
I'll look into it and see if any other problem come around in the future.
My mailman is complied from the source,since my OS X is not a server version.
So I should be on the same boat as most of you guys .:) I'll keep you updated.

Thanks again.

Cen

No. The original error is something else which may or may not have
anything to do with log files, but then scripts/driver attempts to log
the error and can't write the error log, and this is the traceback.


That's because all the log entries you mention are written by
mailmanctl or qrunners which presumably are executing as group
mailman. It is only the web server that can't write the log(s).

Is this Max OS X Server with Apple's mailman or is it your own Mailman
installed from source? If the former, item 8) in
<http://mail.python.org/pipermail/mailman-users/2004-July/038388.html>
might be of interest.