[rabbitmq-discuss] RabbitMQ crashes

Matthew Sackman matthew at rabbitmq.com
Mon Aug 29 11:43:45 BST 2011

Hi Matti,

Well I'm not sure why it's not stopping - I'd have expected it to crash
in a large ball of fire at this point.

On Sun, Aug 28, 2011 at 11:57:11PM -0700, Matti Linnanvuori wrote:
>     exception exit: {{badmatch,eof},
>                      [{rabbit_queue_index,load_segment,2},
>                       {rabbit_queue_index,segment_entries_foldr,3},
>                       {rabbit_queue_index,
>                           '-queue_index_walker_reader/2-lc$^0/1-0-',
> 4},
>                       {rabbit_queue_index,queue_index_walker_reader,
> 2},
>                       {worker_pool_worker,handle_cast,2},

It's actually got seriously upset on start up. It looks like some file
corruption occurred last time it was running or on shut down - it may be
as simple as it ran out of disk space, but equally it may point at other
issues. As a result, on start up, it's trying to read off disk the
various contents of queues, and it's unexpectedly reaching the end of a
file before. Now it's probably a bug that Rabbit doesn't continue
regardless in this situation and certainly the "up but not working"
state that you've reached is not a design goal. Really, at this point,
I'd suggest you kill off rabbit, wipe out its database directory (which
will wipe out all msgs and queues on disk) and then start from a clean
slate. That will lose you all msgs on disk, but at this point, I'm not
sure how much could be recovered anyway.

Can you think of any point at which you may have run out of disk space,
or there's been faulty hardware or file system corruption?


More information about the rabbitmq-discuss mailing list