Reached end-of-stream while waiting for more data from peer

[Damien_Dallimor]

Every now and again I see this message in a distributed search setup with a few peers and a couple of pooled search heads, using mounted bundles.

Oddly enough, it only ever pops up when executing sub searches.I never see the error when using the Search app , but do see it via all other apps. Weird.

It could be many things, so before I embark on diagnosing it , I thought I'd just ask the audience if anyone else has experienced this and/or remedied it ?

DD.

[Damien_Dallimor]

Just an update...I disabled mounted bundles and reverted to bundle replication and things are fine now. This was a Splunk 4.2.4 distributed deployment.

Bit of a head scratcher ,I think what was happening was that when using mounted bundles with distributed sub searches, the inner search was using the mounted bundles path , and then the outer search was using the $SPLUNK_HOME/var/run/searchpeers path , which contained no apps because bundle replication wasn't being used.Deduction based on the output of "Inspect Search Job" under the Actions menu. Quite why the inner and outer searches would not both be using the mounted bundles path is a bit weird ??

Never had any problems with regular searches when using mounted bundles, only sub searches.