indexer-worker panics with latest mercurial

[Timo Sirainen]

On 12 Mar 2015, at 22:55, Laeeth Isharc <laeeth at laeeth.com> wrote:
> 
> Hi.
> 
> I am seeing the following in my logs.  Happy to downgrade to an earlier version if you think this might be the problem, but dovecot.org is extremely slow (and has been for months) so I found it easier just to pull the latest from mercurial.

dovecot.org isn't slow as far as I know, although I've noticed some network connectivity problems to some places .. It's in Azure, so I guess somehow related to that..

> Mar 12 20:48:39 indexer: Error: Indexer worker disconnected, discarding 1 reques
> ts for laeeth at laeeth.com
> Mar 12 20:48:39 indexer-worker(laeeth at laeeth.com): Fatal: master: service(indexe
> r-worker): child 24003 killed with signal 6 (core dumps disabled)
> Mar 12 20:49:01 dsync-local(laeeth at laeeth.com): Error: Couldn't lock /home/mail/
> laeeth_laeeth_com/.dovecot-sync.lock: Timed out after 30 seconds
> Mar 12 20:49:16 indexer-worker(laeeth at laeeth.com): Error: fts_tika: PUT http://l
> ocalhost:9997/tika/ failed: 500 Server Error
> Mar 12 20:49:17 indexer-worker(rosie at kaleidicassociates.com): Warning: I/O leak:
> 0x7fc47d60fcf0 (line 127, fd 25)
> Mar 12 20:49:17 indexer-worker(rosie at kaleidicassociates.com): Panic: file ioloop
> .c: line 39 (io_add_file): assertion failed: (callback != NULL)

I tried this a few times, but I wasn't able to cause the crash. Since you're already pulling from mercurial, can you find which mercurial commit caused it to start breaking?