Rebuilding replicas


(asanderson) #1

Is there a way via API to rebuild a replica other than just restarting the
replica's node? If not, then I will submit a feature request.

Somehow some of our replicas have different doc totals than their
primaries, and I don't see a way to fix it via API.

--
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/9c64d53c-9fe0-412a-8056-b2d17c3ea59f%40googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.


(Paul Smith) #2

You can issue a Move allocation command of just the replicas. A single
move request with each of your replica shards shunted to a different node
should do it. The move is implemented as a copy from the primary (well it
is in ES 0.19 anyway)

Note though that if you move things aroun such that a post move
auto rebalance dictates a shard replica to go back to where it was it
appears to then find the old copy and just reuse that data which cancels
out what you just tried to so. :slight_smile:

Paul

On Thursday, 6 February 2014, asanderson a.steven.anderson@gmail.com
wrote:

Is there a way via API to rebuild a replica other than just restarting the
replica's node? If not, then I will submit a feature request.

Somehow some of our replicas have different doc totals than their
primaries, and I don't see a way to fix it via API.

--
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<javascript:_e(%7B%7D,'cvml','elasticsearch%2Bunsubscribe@googlegroups.com');>
.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/9c64d53c-9fe0-412a-8056-b2d17c3ea59f%40googlegroups.com
.
For more options, visit https://groups.google.com/groups/opt_out.

--
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/CAHfYWB4%3DM%2BFyP1GXKNc_E%3Db4Z08vcOO21p9zcm539KV8L1GnUw%40mail.gmail.com.
For more options, visit https://groups.google.com/groups/opt_out.


(asanderson) #3

You can issue a Move allocation command of just the replicas. A single
move request with each of your replica shards shunted to a different node
should do it. The move is implemented as a copy from the primary (well it
is in ES 0.19 anyway)

Note though that if you move things aroun such that a post move
auto rebalance dictates a shard replica to go back to where it was it
appears to then find the old copy and just reuse that data which cancels
out what you just tried to so. :slight_smile:

Ahhh...so that's the trick. :wink: Well, it definitely wasn't obviously
reading the API guide. Perhaps there should be a note there. :wink:

I actually tried moving the primary which didn't change the replica, but I
confirmed that moving the replica did rebuild it.

MANY 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/61b6aafa-3558-4626-b456-4b71d7b13579%40googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.


(system) #4