Originally from the User Slack
@Daria_Fedorova: Hi, I got an unfamiliar error while repairing keyspace=system on staging scylla cluster. (version is 5.1.19 Т_Т)
stage1 scylla: [shard 0] repair - repair[46ef5754-9e6d-4c4d-8687-96424802c420]: repair_tracker run failed: std::runtime_error ({shard 0: std::runtime_error (All tables repaired together have to have the same sharding logic. Different sharding logic found: sharder[shard_count=78, ignore_msb_bits=12] (for table peer_events) and sharder[shard_count=1, ignore_msb_bits=0] (for table scylla_table_schema_history)),
I did not know the number of shards could be different for tables. How could this happen? Is was misconfigured ? Someone forgot to clean up tables when retrying setup ?
And how do i fix it ?
@avi: You’re not supposed to repair the system keyspace
@Daria_Fedorova:
ok ) forgot it
system_distributed system_schema
Are also skipped when repairing all spaces
But system_traces system_auth system_distributed_everywhere should be repaired ?
@avi: system_traces can be repaired, but it’s not important so better to skip it.
The others must be repaired regularly