Allocation via _id vs _ip

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Allocation via _id vs _ip

Shairon Toledo
Hi guys,

For instability reason(I guess in Amazon EC2) we're moving our cluster from a zone to another, we're doing this node by node. 
Using allocation API by _ip I can exclude a node from cluster. When I include a new node by _ip

  curl -XPUT localhost:9200/_cluster/settings -d '{
      "transient" : {
          "cluster.routing.allocation.include._ip" : "10.34.2.1,10.34.2.2,10.34.2.3,10.34.2.4"
      }
  }'


I see the new node on _cluser/_nodes but shards is not initializing on it even in a period of 3 hours after performed the command, I already tried in transient and persistent mode.

However when I include using _id the shards start initializing right away.

How you experienced something like that? If so, how to troubleshoot that? 

I don't like the idea to use _id for allocate nodes if a node fail and need be restarted, the id become other.


Best,



--











Shairon Toledo
http://hashcode.me

--