Hello
So, first of all, what you did with mongo is quite creative but sadly it won’t work so easily.
Basically you imported Mongo collections which include information about indices YOU DON"T HAVE. Graylog doesn’t know that, Graylog will try to query everything it has information on and it takes information from Mongo collections.
You have to drop some “tables” inside MongoDB, the ones that contain index_ranges for indices. After you do that you have to do what Jan Doberstein told you - recalculate every single index range using Graylog’s “System” => “Indices” => “Recalculate index ranges”.
The hard part is dropping stuff from Mongo, the trival part is recalculating index ranges.
So, dropping stuff in Mongo.
Go into machine you have MongoDB installed on and use those commands, connect using your auth db, example if you were following instructions in docs:
mongo --port 27017 -u “admin” -p --authenticationDatabase “graylog”
Use your password for “admin” user
After connecting to your mongo use those:
use graylog;
← use the graylog’s database, maybe you used a different name?
show collections
← show all collections, you should see a index_ranges collection
db.index_ranges.drop()
← drop all info regarding index_ranges
exit
Okay, you just cleared info regarding those nasty indices you don’t really have in your Elasticsearch cluster but Graylog desperately wanted to query.
Reindexing:
Do what jan wrote, “System” > “Indices” > click on index set > Maintenance > Recalculate index ranges.
Do that on every single index set you have in your Graylog.
That’s it, it should help
Some further info regarding your problem and how to dodge it in the future:
Graylog has to manage your Elasticsearch cluster - every single index creation / deletion has to be made using Graylog. If you delete something from Elasticsearch by some query or Curator or in any other way which is not including clicking “Delete” in Graylog - you will see the error you see right now.
This is why: You delete something manually but information about it is still stored in Mongo so from Graylog’s POV it is NOT DELETED. Same thing for index creation, if you create some indices not using Graylog’s “Create index set” - it won’t work because there will be no info about it in MongoDB/Graylog.
There are multiple operations you can do on indices manually / with Curator / Query, for example you could relocate shard, create custom mappings and so on but deleting / creating indices manually is BIG NO NO