Re: [Mailman-Users] Archival failure (HyperArch?)
I've just set up a 2.0b5 server. I have all the list functionality working right, as far as I know, but there are no HTML archives. The mbox file is being updated -- I can see the messages in there -- but it's not being converted to HTML.
"arch" is being run, though. When I run it manually, I get: Pickling archive state into /var/mailman/archives/private/list/pipermail.pck
And the web stuff says (no matter whether I run arch directly, or let Mailman do it): Currently, there are no archives.
'arch' is not normally run to archive mail; it's merely a "rebuild" convenience command. Mails are normally archived internally by invoking methods in Archiver.py, specifically (usually) ArchiveMail().
However, diagnosing 'arch' would be a good thing to do.
I know I have a weird installation, but I don't know where to begin debugging this -- not enough sense of HyperArch and pipermail's guts. Any pointers?
My standard recommendation: can you run arch under some sort of system-call-trace utility, and watch what files it opens (or tries) and what it tries to write? Solaris: truss Linux: strace
On 2000.08.30, in 200008310456.VAA01628@utopia.west.sun.com, "Dan Mick" Dan.Mick@West.Sun.COM wrote:
'arch' is not normally run to archive mail; it's merely a "rebuild" convenience command. Mails are normally archived internally by invoking methods in Archiver.py, specifically (usually) ArchiveMail().
However, diagnosing 'arch' would be a good thing to do.
OK, that's expected now that I think of it. Still, reword it a little, and the problem's the same.
I know I have a weird installation, but I don't know where to begin debugging this -- not enough sense of HyperArch and pipermail's guts. Any pointers?
My standard recommendation: can you run arch under some sort of system-call-trace utility, and watch what files it opens (or tries) and what it tries to write? Solaris: truss Linux: strace
I could, but as it turns out, that probably won't help. I missed this while looking at the files directly, but following the python line by line suggested that they're not valid mboxes -- and indeed, they were missing the From_.
Actually, that's probably because I'm not using a real local delivery mailer. Unexpected, but understandable.
-- -D. dgc@uchicago.edu NSIT University of Chicago
participants (2)
-
Dan Mick
-
David Champion