On 09/13/2011 11:04 PM, Grzegorz Jaśkiewicz wrote:
> 2011/9/13 Merlin Moncure<mmoncure@gmail.com>:
>> 2011/9/13 Grzegorz Jaśkiewicz<gryzman@gmail.com>:
>>> I'm guessing I won't get much more from devs , without providing more
>>> info here which unfortunately has been lost.
>>
>> yup -- you destroyed all the evidence. if it happens again, try
>> posting some more info, particularly what's going on with locks,
>> pg_stat_activity, maybe an strace of any stuck processes, etc. also
>> irc can be a good place to get some quick help if its time sensitive.
>
> See, I was hoping that just the fact that it occurred might trigger
> some suspicions on the -devel side of things. I know it would in my
> product.
Of course it does. How would _you_ go about reproducing it or tracing it
so you could even begin to track it down, though? There just isn't
enough information to work with.
It's like someone contacting you and explaining that they saved a file
containing a huge embedded image from your office suite software, but
when they went to open the file later it crashed the office suite. They
were in a hurry and didn't have the disk space to keep the old one, so
they had to redo the work and save over the file. It's good to get the
report, but ... what do you do about it? They don't have a faulty file,
a reproducible test case, any or any good prospects for working on
creating a test case.
--
Craig Ringer