Failover for indexing


(Yong Wang) #1

is there amy way to keeping indexing when one node down?


(Christian Dahlqvist) #2

How many nodes do you have in the cluster?


(Yong Wang) #3

9 nodes. The indexing stops when some the node rejoin the cluster.

Wang Yong | Fusionskye


(Christian Dahlqvist) #4

If you have at least 3 master eligible nodes and have minimum_master_nodes correctly set, Elasticsearch should elect a new master in case a node goes down, which should allow indexing to continue.

Which version of Elasticsearch are you on? How much data do you have in the cluster? How many indices/shards do you have? What is the hardware specification of the hosts?


(Yong Wang) #5

Hi, Christian, if one node is down, the search and indexing will survive cause we set number of replica>0.
What I mean is when the dead node is restarted and rejoins the cluster, all shards will be reassigned and re synced. And during this period, the cluster will not index any more document.


(Christian Dahlqvist) #6

How many replicas do you have configured?


(Yong Wang) #7

10P+1R, this is a heavy indexing case that average 40k documents/a.


(Christian Dahlqvist) #8

How long is your cluster in a red state when a node is restarted and joins the cluster? Which version of Elasticsearch are you using? How much data do you have in the cluster? How many indices/shards do you have?


(Yong Wang) #9

Sometimes the cluster will remain in red for about 30minutes. So we have to wait a long time for the status transited from yellow to green. The index request rate is 0 before it comes to green.

We have pretty big indices, about 20 indices, 500G each.
The cluster has 9 nodes, 10P+1R, and the version is 1.7.


(system) #10

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.