[Mailman-Users] bin/upgrade has been broken since2.1.5(orpossibly longer)

[Mark Sapiro]

Knut Auvor Grythe wrote:

>On Fri, Sep 19, 2008 at 08:40:32AM -0700, Mark Sapiro wrote:
>>> Now, I could probably have recovered in some other way, but that's
>>> because I know python. Not all mailman users do, and I suspect they
>>> would have severe difficulties recovering from such a situation.
>> 
>> The recovery is something like:
>> 
>>   find lists/*/request.pck -mtime +365 -delete
>
>In my case, probably. But this doesn't change my point, you just have to
>substitute "python" with "mailman". I certainly wouldn't know whether if
>was safe or not to just delete request.pck files like that. That's why I
>patched the upgrade script instead, since it seemed the safest approach.


Understood. I will probably add something to the UPGRADING document
about this, but I don't know if that would have helped in your case
since I don't know if Debian includes that in their documentation.


>>> Also, even if they actually did notice this problem before the
>>> upgrade, how would they solve it? If I was them I'd probably try to
>>> upgrade to the latest version to see if the crashing stops, only to
>>> see that the upgrade also crashes.
>> 
>> I really don't think at this point (11 releases and over 5 years
>> later) that there are that many people in your situation.
>
>That's your call. I am no longer in that situation, so I'll stop nagging
>you about it :-)


It is a hard call. In a real sense, I haven't fixed the bug I claim to
have fixed. I have fixed the bug that was responsible for setting up
the admindb/upgrade crash, but I haven't fixed the crash. On the other
hand, the relative invisibility of this issue over the last 5+ years
tells me your situation is rare. I'll continue to think about it.


>> BTW, it seems that given your situation, you may have a large number of
>> dead held messages. If you haven't done so, you might look at the FAQ
>> at <http://wiki.list.org/x/nIA9> and remove them.
>
>That is indeed relevant in my situation. However, I just set
>max_days_to_hold to 60 for all lists. Shouldn't that take care of it
>automatically?


Yes, that will take care of it.

-- 
Mark Sapiro <mark at msapiro.net>        The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan