Safe to ignore a {sink_died, normal} error?

Martin Woods mw2134 at gmail.com
Wed Mar 14 07:00:27 EDT 2012


Hi

We're using Riak 1.1.1 (on Ubuntu, built from source) and are seeing a
number of errors raised regularly in the error logs during mapreduce. I
think I remember reading a post on this forum a while ago that these error
messages are nothing to worry about, that they are just a result of one
aspect of the mapreduce processing shutting down before receiving a normal
shutdown message from another process - hence the {sink_died, normal}. The
comments in the source code where this error is generated seem to back this
up. Indeed, the mapreduce operations in question all completed just fine.

Can someone from Basho please confirm that this is the case and that we're
safe to ignore them? Here are the entries in error.log:

2012-03-13 23:25:12.810 [error] <0.10984.0> gen_fsm <0.10984.0> in state
wait_pipeline_shutdown terminated with reason: {sink_died,normal}
2012-03-13 23:25:12.945 [error] <0.10984.0> CRASH REPORT Process
<0.10984.0> with 0 neighbours crashed with reason: {sink_died,normal}
2012-03-13 23:25:12.989 [error] <0.165.0> Supervisor riak_pipe_builder_sup
had child undefined started with {riak_pipe_builder,start_link,undefin
ed} at <0.10984.0> exit with reason {sink_died,normal} in context
child_terminated

Thanks and regards,
Martin.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.basho.com/pipermail/riak-users_lists.basho.com/attachments/20120314/96663b7f/attachment.html>


More information about the riak-users mailing list