Currently, restoring red indices from a snapshot require the following steps in Kibana UI:
1) Close all the red indices
Go to Stack Management > Index Management
Toggle "Include hidden indices"
Sort by Health column or search for: health:red
At the bottom of the page select: Rows per page: 100 if you have a high number of red indices that need to be closed
Select all red indices checkboxes
Click on Manage x indices > Close indices
Save the list of red indices somewhere (i.e. text editor). This is a manual process which require to copy-paste indices from the Index Management or the usage of APIs in Kibana Dev Tools (e.g GET _cat/indices?health=red&h=index)
(You may need to repeat the operation if you have more than 100 red indices as there is a limit of 100 rows per page)
2) Restore the closed red indices
Go to Stack Mangement > Snapshot and Restore
Select the latest successful snapshot (State: Snapshot Complete) from the list
(i.e Snapshot with most recent Date Created and no Failed Shards)
Click Restore
Toggle All data streams and indices
Click Select data streams and indices
Select the red indices from the dropdown (by cross-checking from the list you saved in 1. which can be very time-consuming and prone to errors)
Toggle off Restore aliases (as often not needed for red indices recovery scenarios)
Next, Next and Restore snapshot
Describe the feature
Add a feature to automate and streamline the restore of red indices in a cluster.
Proposals
Below, 2 different alternatives that could be implemented as a feature:
1. Only restore red indices check box from the Restore UI
2. Restore from Index Management
Filter and select the desired indices. Then, Restore indices should prompt a message which allow the users to select from which repository / snapshot they want to restore from and populate the restore UI accordingly (with the red indices pre-selected).
In both proposals (1) and (2), the step of closing red indices to restore should be tranparent for the user - i.e operated in the background before the restore happens.
I'm open to discuss other solutions for this enhancement in order to make the restore process of red indices more user-friendly.
CC: @romain-chanu who participated in the brainstorming :heart:
Kibana version: 8.3.2 Related issues:
Synopsis
Currently, restoring red indices from a snapshot require the following steps in Kibana UI:
1) Close all the red indices
health:red
Rows per page: 100
if you have a high number of red indices that need to be closedGET _cat/indices?health=red&h=index
)(You may need to repeat the operation if you have more than 100 red indices as there is a limit of 100 rows per page)
2) Restore the closed red indices
State: Snapshot Complete
) from the list (i.e Snapshot with most recentDate Created
and noFailed Shards
)Describe the feature
Add a feature to automate and streamline the restore of red indices in a cluster.
Proposals
Below, 2 different alternatives that could be implemented as a feature:
1. Only restore red indices check box from the Restore UI
2. Restore from Index Management
Filter and select the desired indices. Then, Restore indices should prompt a message which allow the users to select from which repository / snapshot they want to restore from and populate the restore UI accordingly (with the red indices pre-selected).
In both proposals (1) and (2), the step of closing red indices to restore should be tranparent for the user - i.e operated in the background before the restore happens.
I'm open to discuss other solutions for this enhancement in order to make the restore process of red indices more user-friendly.
CC: @romain-chanu who participated in the brainstorming :heart: