-
Notifications
You must be signed in to change notification settings - Fork 45
Cannot create policy after upgrading to v1.12 #388
Comments
Hi @tocw, Can you paste the mappings of your |
We ran into this same issue when upgrading to 1.12, we had been using our own fork of the ISM plugin with the allocation action as implemented in I believe #106 before it was officially merged. It's unclear to me whether this code changed in the final release, but in any case, we were able to resolve it with the following steps, most easily performed in the kibana dev console:
Note that you probably need to get rid of the top-level key ( |
Hello @dbbaughe , here are the mappings:
|
Hello @samling , |
Hi @tocw, based on your mappings it does seem like you used a fork with the Allocation PR before it was merged in? You can always add more mappings to an index, but you can't modify existing mappings which unfortunately is what you need to do as you have mappings for allocation that don't exist in the plugin and conflict with the plugin's mappings. You will have to ultimately do some variation of what @samling posted to get back in line with our production release of the plugin. You'll have to make copies of whats in the config index, delete the existing one, and then add the copies back into the config index ensuring you have the correct mappings. You can just create a dummy policy when you want to recreate the index and the plugin will create it all with the correct mappings. The issue you'll run into though is by deleting the index and the jobs the plugin will delete the metadata in the cluster state. You might need to make a bunch of temporary policies that you can re-apply to existing indices that are at different states of their lifecycles to get around it. |
Hi @dbbaughe ,
What do you think about it? |
Hello,
after upgrading to 1.12 I noticed we are not able to create ISM policies.
I simply took the example from your documentation here and this is the error message I got:
We run pure 1.12 release with no additional modifications. Please investigate as this is a critical issue for our production clusters
The text was updated successfully, but these errors were encountered: