Occasional scroll timeout before the given timeout value
This post has NOT been accepted by the mailing list yet.
I'm performing a scroll query to scroll through results. Start to end, the process takes less than 5 seconds.
I've tried various scroll timeouts on my scroll query (currently set to "10m"), but it doesn't stop the fact that occasionally I'll get a scroll timeout error. The operation will still take no longer than 10 seconds.
My scroll timeout value seems to be being ignored.
Is there a limit on the length of scroll timeout? ie: If I set mine to 10m, as above, will there be something somewhere saying: "Noooo...that's too long. We'll automatically reduce it to 5s" or something?!
Basically, why is this happening? It doesn't happen ALL the time. Just every now and again for about a minute, and the the problem clears itself and it.
It's worth mentioning that I'm aggressively indexing at the same time.
Could it be that the connection count has been met by the indexer and that subsequent operations are being treated as timeouts instead of no "node available"s or similar?
This last suggestion would explain why sometimes it happens for a short period and then clears out again...