Open mikliapko opened 1 month ago
@karol-kokoszka @Michal-Leszczynski Guys, could you please review and spare your thoughts about the tests?
Test 1 is really useful as we don't cover mixed setup with our own test right now.
I'm not sure about correctness of Test 2. In theory SM requires that backed-up cluster and restore destination cluster have the same schema and this test scenario breaks this assumption. It would be an interesting experiment, but it's not something that is currently covered by SM functionality.
@mikliapko I agree with Michał. Test 2 is against the restore pre-requisite which is to have the same schema -> https://manager.docs.scylladb.com/stable/restore/restore-tables.html#prerequisites
In general, tablet migration can happen at anytime and it could make SM skip some tablets when performing repair. SM should stop tablet migration when performing repair and resume it afterwards.
This gives an idea for additional test scenario:
initial_tablets
keyspace/table option would work)@mikliapko please open an issue to scylla OSS to have an endpoint to check the status of tablet migration (ongoing or not). Mention this issue here so we will have the link. Thanks
A few tests can be introduced here:
Test 1 Run manager sanity checks against the database with two keyspaces:
Test 2