[Mailman-Developers] Cron error messages

[bronto]

Thanks for the reply, and yes, there was a hard disk failure a couple 
weeks prior to this.  Perhaps corrupted python/mailman was the first 
effect, that didn't show up until later.

Since the original "bad marshal data" problem, I've upgraded to 
v2.1b3, so unless make install does not replace existing files I 
think my mailman installation should be OK now (?).  I will 
re-install python2 to make sure it's OK as well.

The remaining question is that of all the files in the shunt 
directory.  These must represent missing posts from the mailing list. 
Can these be recovered and processed?  Is it simply a matter of 
moving them into a different directory?  To me, "shunt" sounds like 
"move these here because I can't deal with them".

Thanks again

Rob



>  >>>>> "bronto" ==   <bronto@csd-bes.net> writes:
>
>     | Traceback (most recent call last):
>     |    File "/usr/local/mailman/cron/gate_news", line 44, in ?
>     |      from Mailman import MailList
>     | ValueError: bad marshal data
>
>Because you're getting this on an import, I have to conclude that your
>.pyc files are corrupt.  I've no idea why that could be (maybe you ran
>out of disk space once, or have other h/w problems).
>
>One thing you can do is to temporarily disable cron, the web
>interface, and qrunner, then cd to /usr/local/mailman and run this:
>
>% find . -name \*.pyc -print | xargs rm
>
>Then cd back to your source directory and do a "make install".  This
>will regenerate all your pyc files.  Watch carefully for any errors,
>especially during the "compileall" phase.
>
>     bronto> Since posting that, I have discovered that one of my
>     bronto> mailing lists has stopped functioning to some degree.  I
>     bronto> had set up this particular list to be a simple personal
>     bronto> archive of messages from another mailing list I subscribe
>     bronto> to.  It doesn't send anything, so I hadn't noticed any
>     bronto> problems.  While doing my mailman update to 2.1b3, I
>     bronto> discovered that no messages have been archived since the
>     bronto> 23rd when I got the above errors.  Snooping around the
>     bronto> mailman directories, I see a ton of files (several
>     bronto> hundred) in the qfiles/shunt directory, dated since 9/23.
>     bronto> There are also a few dozen from 7/2, FWIW.  I'm betting
>     bronto> this isn't a coincidence, but know nothing more than that.
>
>     bronto> Today at 12 noon, I received the following cron error:
>
>     | _________________________________________
>     | Traceback (most recent call last):
>     |    File "/usr/local/mailman/cron/gate_news", line 44, in ?
>     |      from Mailman import MailList
>     |    File "/usr/local/mailman/Mailman/MailList.py", line 31, in ?
>     |      import urllib
>     | EOFError: EOF read where object expected
>     | __________________________________________
>
>Again, that's happening on an import, but now it's while importing a
>standard Python module.  That might mean that your Python installation
>is corrupt in a similar way.
>
>     bronto> Since that time, there has been a mailman process running
>     bronto> taking up all available CPU cycles.  gtop shows that the
>     bronto> CMD is /usr/bin/python2, FWIW.  There are a fair number of
>     bronto> qrunner processes running, but my mailing lists are hardly
>     bronto> active.
>
>     bronto> What's going on here?
>
>I don't know if those other problems are related, but you first have
>to make sure those import problems go away before you can track down
>any other issues.
>
>-Barry