We seem to have found it; the user_id biz was a red
herring.
The stuff in Burner.pm to
# Expire stale resources, if necessary.
queries Bric: ist::Res
ource with 'not_job_id' => $thisJobId
and that query just seems to hang here. I rewrote that query
in
Resource.pm with a second subquery
AND r.id NOT IN (
SELECT resource__id
FROM job__resource
WHERE " . any_where($jid, 'job__id = ? ', params) .
"
)
and that seemed to fix it.
Submitted http://
bugs.bricolage.cc/show_bug.cgi?id=1268 with patch.
--Brad
john grumpet.net wrote:
> Hello,
>
> We're having an issue with publishing media - it seems
slightly
> related to an issue raised on this list earlier in the
month, but also
> different in a significant way. A bit of background
that may be
> useful - this instance of bric (1.10.2) was recovered
from a disk
> failure in an unorthadox way; the files were restored
from a backup,
> and the db was already on a separate machine, so they
were just
> reunited. I doubt this is important, but you never
know!
>
> Anyway, I can preview an image successfully, but
publishing hangs
> indefinitely. Same goes for trying via bric_soap with
an unlimited
> timeout. Generally, nothing makes it to the log, so
there's no useful
> debug info. From some of my investigation, it would
appear it dies
> around where it's trying to get a user_id.
>
> Stories can publish fine.
>
> The postgres version is 7.4.6
>
> Any thoughts or suggestions would be welcome - We could
use a fresh
> perspective!
>
> Thanks,
>
> John
>
>
|