ISM error prevention resolutions

Resolutions of errors for each validation rule action are listed in the following sections.

Table of contents

1. The index is not the write index
2. The index does not have an alias
3. Skipping rollover action is true
4. This index has already been rolled over successfully
5. The rollover policy misses rollover_alias index setting
6. Data too large and exceeding the threshold
7. Maximum shards exceeded
8. The index is a write index for some data stream
9. The index is blocked
10. Remote store is not enabled
11. Segment replication is not enabled
12. No search replicas configured

The index is not the write index

To confirm that the index is a write index, run the following request:

GET <index>/_alias?pretty

If the response does not contain "is_write_index" : true, the index is not a write index. The following example confirms that the index is a write index:

{
  "<index>" : {
    "aliases" : {
      "<index_alias>" : { 
        "is_write_index" : true
      }
    }
  }
}

To set the index as a write index, run the following request:

PUT <index>
{
  "aliases": {
    "<index_alias>" : {
        "is_write_index" : true
    }
  }
}

The index does not have an alias

If the index does not have an alias, you can add one by running the following request:

POST _aliases
{
  "actions": [
    {
      "add": {
        "index": "<target_index>",
        "alias": "<index_alias>"
      }
    }
  ]
}

Skipping rollover action is true

In the event that skipping a rollover action occurs, run the following request:

 GET <target_index>/_settings?pretty

If you receive the response in the first example, you can reset it by running the request in the second example:

{
  "index": {
    "opendistro.index_state_management.rollover_skip": true
  }
}

PUT <target_index>/_settings
{
    "index": {
      "index_state_management.rollover_skip": false
  }
}

This index has already been rolled over successfully

Remove the rollover policy from the index to prevent this error from reoccurring.

The rollover policy misses rollover_alias index setting

Add a rollover_alias index setting to the rollover policy to resolve this issue. Run the following request:

PUT _index_template/ism_rollover
{
  "index_patterns": ["<index_patterns_in_rollover_policy>"],
  "template": {
   "settings": {
    "plugins.index_state_management.rollover_alias": "<rollover_alias>"
   }
 }
}

Data too large and exceeding the threshold

Check the JVM information and increase the heap memory.

Maximum shards exceeded

The shard limit per node, or per index, causes this issue to occur. Check whether there is a total_shards_per_node limit by running the following request:

GET /_cluster/settings

If the response contains total_shards_per_node, increase its value temporarily by running the following request:

PUT _cluster/settings
{
   "transient":{
      "cluster.routing.allocation.total_shards_per_node":100
   }
}

To check whether there is a shard limit for an index, run the following request:

GET <index>/_settings/index.routing-

If the response contains the setting in the first example, increase its value or set it to -1 for unlimited shards, as shown in the second example:

"index" : {
        "routing" : {
          "allocation" : {
            "total_shards_per_node" : "10"
          }
        }
      }

PUT <index>/_settings
{"index.routing.allocation.total_shards_per_node":-1}

The index is a write index for some data stream

If you still want to delete the index, check your data stream settings and change the write index.

The index is blocked

Generally, the index is blocked because disk usage has exceeded the flood-stage watermark and the index has a read-only-allow-delete block. To resolve this issue, you can:

1. Remove the -index.blocks.read_only_allow_delete- parameter.
2. Temporarily increase the disk watermarks.
3. Temporarily disable the disk allocation threshold.

To prevent the issue from reoccurring, it is better to reduce the usage of the disk by increasing disk space, adding new nodes, or removing data or indexes that are no longer needed.

Remove -index.blocks.read_only_allow_delete- by running the following request:

PUT <index>/_settings
{
    "index.blocks.read_only_allow_delete": null
}

Increase the low disk watermarks by running the following request:

PUT _cluster/settings
{
    "transient": {
        "cluster": {
            "routing": {
                "allocation": {
                    "disk": {
                        "watermark": {
                            "low": "25.0gb"
                        }
                    }
                }
            }
        }
    }
}

Disable the disk allocation threshold by running the following request:

PUT _cluster/settings
{
    "transient": {
        "cluster": {
            "routing": {
                "allocation": {
                    "disk": {
                        "threshold_enabled" : false
                    }
                }
            }
        }
    }
}

Remote store is not enabled

The search_only action requires remote store to be enabled on the cluster. Remote store must be enabled at cluster creation time and cannot be enabled on an existing cluster. For more information, see Remote-backed storage.

Segment replication is not enabled

The search_only action requires segment replication to be enabled for the index. Segment replication must be configured at index creation time. For more information, see Segment replication.

No search replicas configured

The search_only action requires at least one search replica. For more information about configuring search replicas, see Separate index and search workloads.