When you scale out or scale in a GBase 8a mirror cluster, the operation changes the number of primary shards (hashmaps) or the distribution count. This immediately breaks the mirror prerequisite: both sides must have the same number of primary shards and compatible distributions. As a result, the rebalance process gets stuck indefinitely and cannot finish. Here is how to resolve it in your gbase database.
The Symptom
The gcluster express.log shows an error like this:
<RebalanceRunQuery|259>:gcluster command error: vc00001's hashmap number(2) do not equal vc00002's hashmap number(1)
The rebalance operation simply cannot proceed because the mirror relationship is now invalid.
The Fix
Remove the mirror relationship first. Once the expansion is complete and the hashmaps are aligned again, you can re‑establish the mirror.
ALTER DATABASE testdb DELETE MIRROR;
After running this command, the rebalance will continue and finish normally. This is the standard approach whenever a topology change invalidates a mirror configuration in GBASE's MPP cluster. Always verify that both sides have matching hashmap counts before re‑creating the mirror to keep your gbase database synchronisation healthy.
Top comments (0)