Cluster Routing Allocation Settings

[xavier.tiendeo (xavi)]

We have a cluster of two nodes (ES 2.2), indexes have one replica so all shards are in both nodes. If we restart a node, all shards in that node are set as unasigned, we think this can be resolved with

"transient": {
"cluster": {
"routing": {
"allocation": {
"enable": "none"

but we don't understand the meaning of another parameter disable_allocation": "false"? is from another version?

how can then reset transient settings?

thanks

[warkolm (Mark Walkom)]

@xavier.tiendeo

If we restart a node, all shards in that node are set as unasigned, we think this can be resolved with

It won't prevent that, you will still have half of your shards unassigned.

@xavier.tiendeo

but we don't understand the meaning of another parameter disable_allocation": "false"? is from another version?

It's the same thing, just an older variant/

@xavier.tiendeo

how can then reset transient settings?

That's difficult until 5.X, so you should upgrade.

[xavier.tiendeo (xavi)]

Thansk for the response, can we prevent " all shards in that node are set as unasigned" in another way?
so we have no shards unassigned, or at least a big % of them?

[warkolm (Mark Walkom)]

Don't shut down the node, or add more nodes.