Question about Rebalancing and data loss

Hi ,

We have recently started using elasticsearch and I have very basic question
about how rebalancing works between nodes if shards are being moved due to
a failure.

eg.

Lets say we have ES deployed in AWS Ec2 region and replicas are "Zone
Aware" and num of replicas = 1.

Considering that we have 3 zones in Ec2, if One entire zone goes down, ES
will start rebalancing shards associated with failed zone to remaining Two
zones.

And before rebalancing is complete (entire data set is not yet copied), if
failed zone comes back up again, does ES try to rebalance (or move) shards
which have partial data to the new zone ?

or does it wait for previous rebalancing to finish (until entire data set
is copied) and then start rebalancing again ?

If answer is former then will there be any data loss due to moving shards
with partial data ?

Thanks,

--
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/e33f1b7b-d27a-48dd-8b55-d101bc2cd779%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Does anybody know about this ?

Thanks,

On Thursday, March 27, 2014 4:22:47 PM UTC-7, sagarl wrote:

Hi ,

We have recently started using elasticsearch and I have very basic
question about how rebalancing works between nodes if shards are being
moved due to a failure.

eg.

Lets say we have ES deployed in AWS Ec2 region and replicas are "Zone
Aware" and num of replicas = 1.

Considering that we have 3 zones in Ec2, if One entire zone goes down, ES
will start rebalancing shards associated with failed zone to remaining Two
zones.

And before rebalancing is complete (entire data set is not yet copied), if
failed zone comes back up again, does ES try to rebalance (or move) shards
which have partial data to the new zone ?

or does it wait for previous rebalancing to finish (until entire data set
is copied) and then start rebalancing again ?

If answer is former then will there be any data loss due to moving shards
with partial data ?

Thanks,

--
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/a3074a59-9b3c-48af-b28d-6fa977e6ed7b%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Replicas will be recovered to meet your replica count requirements. If a
node comes down, and then recovery happens, and then the node comes back,
recovery will run until it is complete after which a rebalance may happen.
No data loss will happen at any point.

--
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/8b64dd27-e9c5-40d9-9abb-cef9bf78c6c3%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.