Still preparing an migration from elastic to opensearch. run some testing and it seems the need to bring our dataminer cluster down for approx 8 hours to just reindex the database.
two questions:
1) can i delete indices not being just anymore without issue? (for instance the Job manager ones: dms-cjobsectiondefinition, and for instance suggestions alarm older like dms-suggest-alarms-2021.07.30.05-000013 for instance.)
2) since monitoring/data loss will be inevitable (dataminer down for 8 hoursa). can't i just start the migration process, then cut over to opensearch meaning that dataminer will go back to the data from the moment the snapshot is taken. Or will this cause issues with the cassandra database for instance?
Hi Gerwin,
I can affirmatively answer both questions.
Removing the older unused and suggestion indices is safe to do for this migration and can shave off some time off the process.
You can run the reindexing in parallel with a running system, the main thing to note is that any configuration that is saved in the indexing database over the course of the process will be lost. This mainly maters for DOM, SRM and alarms as far as the indexing database is concerned.
If that data loss is acceptable then you can certainly keep the system up and running whilst re-indexing.
With Kind Regards,