Hi Everyone,
Greetings…!!
If we have a cluster setup of 2 aggregators and 6 leafs…and if we lose 2 leaf nodes from 1 availability group (1) and 1 leaf node from the availability group (2). Will my cluster still be available for the read/write operations or will it go down.?
It depends. If pair leaf nodes are both down then data won’t be available. If not, it will be available.
Check for the pair host and port using the view mentioned in this doc: LEAVES · SingleStore Documentation
Thanks
In the case where it is not a pair that goes down, will the cluster automatically move to the nodes in the other AG or are there manual steps required to do so? Assuming the downed nodes were temporary, once returned to service, will the pair automatically synch the data?
In a SingleStore cluster configured for high availability, if a leaf node goes down, the cluster will automatically failover partitions to its paired node within the configured availability group (AG), if in ‘paired’ mode, or rebalance across the cluster, if in ‘load_balanced’ mode. This behavior occurs without manual intervention due to the cluster’s high availability features.
Once a downed node returns to service, it transitions from the ‘offline’ to the ‘attaching’ state. During this phase, the master aggregator detects the node and automatically runs ATTACH LEAF to reintroduce the leaf’s partitions back into the cluster. The system automatically attempts to reuse the data present on the leaf, initiating a process to synchronize and recover any changes missed while the node was down.
Refer the doc links.