honeymoose
/
OpenSearch
mirror of https://github.com/honeymoose/OpenSearch.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Default delayed allocation timeout to 1m from 0 

Change the default delayed allocation timeout from 0 (no delayed allocation) to 1m. The value came from a test of having a node with 50 shards being indexed into (so beefy translog requiring flush on shutdown), then shutting it down and starting it back up and waiting for it to join the cluster. This took, on a slow machine, about 30s.
The value is conservatively low and does not try to address a virtual machine / OS restart for now, in order to not have the affect of node going away and users being concerned that shards are not being allocated to the rest of the cluster as a result of that. The setting can always be changed in order to increase the delayed allocation if needed.
closes #12166
This commit is contained in:
Shay Banon 2015-07-09 19:12:07 +02:00
parent 7fec35b678
commit e598f16b58
3 changed files with 6 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
core/src/main/java/org/elasticsearch/cluster/routing/UnassignedInfo.java
View File

@ -42,7 +42,7 @@ public class UnassignedInfo implements ToXContent, Writeable<UnassignedInfo> {
public static final FormatDateTimeFormatter DATE_TIME_FORMATTER = Joda.forPattern("dateOptionalTime");
public static final String INDEX_DELAYED_NODE_LEFT_TIMEOUT_SETTING = "index.unassigned.node_left.delayed_timeout";
private static final TimeValue DEFAULT_DELAYED_NODE_LEFT_TIMEOUT = TimeValue.timeValueMillis(0);
private static final TimeValue DEFAULT_DELAYED_NODE_LEFT_TIMEOUT = TimeValue.timeValueMinutes(1);
/**
* Reason why the shard is in unassigned state.

4
core/src/test/java/org/elasticsearch/test/InternalTestCluster.java
View File

@ -50,6 +50,7 @@ import org.elasticsearch.cluster.node.DiscoveryNode;
import org.elasticsearch.cluster.node.DiscoveryNodes;
import org.elasticsearch.cluster.routing.OperationRouting;
import org.elasticsearch.cluster.routing.ShardRouting;
import org.elasticsearch.cluster.routing.UnassignedInfo;
import org.elasticsearch.cluster.routing.allocation.decider.DiskThresholdDecider;
import org.elasticsearch.cluster.routing.allocation.decider.ThrottlingAllocationDecider;
import org.elasticsearch.common.Nullable;
@ -482,6 +483,9 @@ public final class InternalTestCluster extends TestCluster {
builder.put(ScriptService.SCRIPT_CACHE_EXPIRE_SETTING, TimeValue.timeValueMillis(RandomInts.randomIntBetween(random, 750, 10000000)));
}
// always default delayed allocation to 0 to make sure we have tests are not delayed
builder.put(UnassignedInfo.INDEX_DELAYED_NODE_LEFT_TIMEOUT_SETTING, 0);
return builder.build();
}

2
docs/reference/index-modules/allocation/delayed.asciidoc
View File

@ -32,7 +32,7 @@ requests) which have been automatically <<indices-synced-flush,sync-flushed>>.
The allocation of replica shards which become unassigned because a node has
left can be delayed with the `index.unassigned.node_left.delayed_timeout`
dynamic setting, which defaults to `0` (reassign shards immediately).
dynamic setting, which defaults to `1m`.
This setting can be updated on a live index (or on all indices):