out_norikra doesn't retry to prepare any targets after the thread failed

[yoheimuta]

Sometimes (a few times a day in our system) I encountered the failure to prepare a target with error messages like below.

2015-04-01 23:19:47 +0900 [error]: norikra/output.rb:131:rescue in prepare_target: failed to prepare target:stdout_ad norikra="norikra1.mtburn.internal:26571" error=Errno::ETIMEDOUT message="Connection timed out - connect(2) for \"norikra1.mtburn.internal\" port 26571 (http://norikra1.mtburn.internal:26571)"
2015-04-01 23:19:47 +0900 [error]: norikra/output.rb:99:block in register_worker: Failed to prepare norikra data for target:stdout_ad

I could simulated this situation that the thread to run register_worker stopped after the same failure.

• Because @norikra_started.push(t) throw NoMethodError.

This line has remained from v0.0.1, so I supposed this exception means something important.

But I want to avoid this failure (and subsequent forever RuntimeError until td-agent restart).

Is there any idea to avoid this failure ?

#6 is a similar issue about result except for type.

[tagomoris]

Oh, this is bogus bug...

[yoheimuta]

Thank you for reply. I pulled request #8.

[tagomoris]

I fixed this bug at a809183 and released v0.3.0. Targets should be retried.
Thank you for reporting!

[yoheimuta]

Thanks !