We are seeing a slow down in shard initialization speed as the number of
shards/indices grows in our cluster.
With 0-100's of indices/shards existing in the cluster a new bulk creation
of indices up the 100's at a time is fine, we see them pass through the
states and get a green cluster in a
Sounds like the inevitable add more nodes situation.
How much RAM on each node, how big is your data set?
Regards,
Mark Walkom
Infrastructure Engineer
Campaign Monitor
email: ma...@campaignmonitor.com
web: www.campaignmonitor.com
On 13 May 2014 19:59, Paul codive...@gmail.com wrote:
We are
In testing and replicating the issue, this slow down has been seen
occurring with empty indices.
The running cluster is at present ~100 GB across 2,200 Indices with a total
of 13,500 shards and ~430,000,000 documents.
We have 7GB RAM and 5GB heap on the data nodes - haven't looked overly
Empty or not, there is still metadata that ES needs to maintain in the
cluster state. So the more indexes you have open the bigger that is and the
more resources required to track it.
Regards,
Mark Walkom
Infrastructure Engineer
Campaign Monitor
email: ma...@campaignmonitor.com
web:
Ok, do you know if there are clear indicators when limits are being reached?
We don't see errors in the logs (apart from the 30s timeout) but if there
are system or ES provided metrics that we can track to know when we need to
scale it would be really useful.
Thanks,
Paul.
On Tuesday,
You will want to obtain Marvel (
http://www.elasticsearch.org/guide/en/marvel/current/) and then wait till
you have a history and start digging.
Regards,
Mark Walkom
Infrastructure Engineer
Campaign Monitor
email: ma...@campaignmonitor.com
web: www.campaignmonitor.com
On 13 May 2014 20:29,
This API should give an indication on any backlog in processing the cluster
state:
http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/cluster-pending.html
On Tuesday, May 13, 2014 11:29:20 AM UTC+1, Paul wrote:
Ok, do you know if there are clear indicators when limits are
Thanks Mark, we'll have a look at the available metrics.
On Tuesday, May 13, 2014 11:34:51 AM UTC+1, Mark Walkom wrote:
You will want to obtain Marvel (
http://www.elasticsearch.org/guide/en/marvel/current/) and then wait till
you have a history and start digging.
Regards,
Mark Walkom
This looks very interesting, thanks.
On Tuesday, May 13, 2014 11:38:27 AM UTC+1, Mark Harwood wrote:
This API should give an indication on any backlog in processing the
cluster state:
http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/cluster-pending.html
On Tuesday,
You should create indexes before bulk indexing. First, bulk indexing works
much better if all indices and their mappings are already present, the
operations will run faster and without conflicts, and the cluster state
updates are less frequent which reduces some noise and hiccups. Second,
setting
Thanks Jörg, we've heard of others pre-creating indices, we were seeing it
as a work around rather than a regular action but what you say makes it
seem like something we should work with.
On Tuesday, May 13, 2014 12:13:10 PM UTC+1, Jörg Prante wrote:
You should create indexes before bulk
11 matches
Mail list logo