File Store Fills Up Doesn't Tell the Downstream Scribe Processes to Buffer Locally

[TomTom]

Hey Fellows,
Have you guys ever seen this before?
If the end point Scribe File Store's file system fills up, it doesn't
tell the downstream (midtier, client) scribe processes to buffer
locally. This causes Scribe messages to be permanently lost.

Has anyone encountered this before? If my scribe aggregator were to
lose both its primary and secondary file stores, I would expect the
down stream scribe servers to buffer to local disk until the Scribe
aggregator came back on line.

Any help with this problem, would be greatly appreciated. Please let
me know if a misconfiguration jumps out at anyone, or if there are any
suggestions, or known fixes

Have a great night!
-Tom

[Anthony Giardullo]

I had added an option called "must_succeed" to deal with this
particular case when the entire buffer store fails. I didn't
originally make it the default behavior to maintain backwards
compatibility. But after thinking about it, this should clearly be
the default behavior. I will make this change and post a note to this
group.

Thanks,
Anthony

[TomTom]

Thanks Anthony,
Your the Man!
Is "must_succeed" a config file option? I couldn't find it. Would I
specify must_succeed=yes or must_succeed=true.
Should I specify it on each of the the three types of scribe servers,
or just on the end point log aggregator.

Thanks a lot for the help!
-Tom

[Anthony Giardullo]

If you download the latest Scribe code, this option is now on by
default and you do not have to change anything.

Otherwise, you just need to specify 'must_succeed=yes' in each store.
See the configuration options on the github wiki.

-Anthony