You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
What is the bug?
I'm unable to restore an index to OpenSearch from a Snapshot where the Snapshot was taken at the time of the index being the write index for the alias.
Since the snapshot was taken, ISM has naturally rolled over and created new write indices.
I have an ISM enabled index, continually rolling over and deleting indices as they age out. During this process I take a snapshot and store it in my defined repository.
How can one reproduce the bug?
Steps to reproduce the behaviour:
Configure an ISM enabled index, rolling over at regular intervals
Take a snapshot and wait for the ISM policy to cycle through, rolling over and deleting the index
Attempt to restore the index that was (at the time of the snapshot) the write index of an alias
See error "Failed to restore snapshot", "alias [rollover_alias] has more than one write index "
What is the expected behaviour?
I should be able to restore the chosen index, setting explicitly that it should not be the write index.
Maybe worth opening an issue in opensearch core to allow restoring an index without the write alias or without specific aliases. There is a setting to “include_aliases” that can be set to false but it affects both read and write aliases so it would cause unexpected behavior. We might need a “include_write_aliases” that will restore without the write aliases of the indices in the snapshot. This would in turn be used by the index management plugin policy.
@opensearch-project/admin This seems to be an OpenSearch core issue since both rollover and snapshot actions are OpenSearch core actions. Please move this issue to OpenSearch core repository.
What is the bug?
I'm unable to restore an index to OpenSearch from a Snapshot where the Snapshot was taken at the time of the index being the write index for the alias.
Since the snapshot was taken, ISM has naturally rolled over and created new write indices.
I have an ISM enabled index, continually rolling over and deleting indices as they age out. During this process I take a snapshot and store it in my defined repository.
How can one reproduce the bug?
Steps to reproduce the behaviour:
What is the expected behaviour?
I should be able to restore the chosen index, setting explicitly that it should not be the write index.
What is your host/environment?
Do you have any screenshots?
Screenshot attached.
Issue initially raised on the OpenSeach Slack channel;
https://opensearch.slack.com/archives/C051JEH8MNU/p1727277154286109
The text was updated successfully, but these errors were encountered: