Manage priority for recovery/index/search


(Павел Суслов) #1

Hello,

My cluster consists of 20 nodes (3 masters + 17 data nodes) handles apr.
100 indices (8 shards + 1 replica). At the moment there is no client nodes,
all bulk requests balancing between 3 master nodes, all search requests
balancing between all 20 nodes. Application can close "cold" indices and
open indices on demand. Most of time only 8 indices are opened. During
dynamic open/close indices (recovery, rebalance processes) I see
performance degradation of searching and bulk indexing in currently active
(green) indices. Normal bulk index request takes less than 1 second, but
while recovering or relocating other indices one bulk request can take more
than 100 sec.
As I understand, recovery and rebalancing have a higher priority than bulk
and searching. Is there any possibility to manage priority for recovery and
rebalancing for indices that open on demand? Or how can I reduce
search/index degradation while recovering other indices. Application allows
only requests to green indices.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/ab345b83-f75e-4ce9-8e3e-8d955a403f31%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.


(system) #2