Hello Eugene, 1) Split brain resolver takes into account only server nodes (not client). No difference between in-memory only or with persistence. 2) It's no necessary to immediately remove a node from baseline topology after split-brain. If you lost backup factor for some partitions (All partition owner primary/backup are in shut down part of a cluster) and explicitly change shrink baseline topology, your affinity for all partitions will be changed and you will lose the data contained in shut down nodes. When split-brain have resolved you may up shutting-down nodes and join them to existing baseline topology, after that outdated data will re-balanced. In the conclusion, you should carefully operate with baseline and do not change it without necessary.
сб, 15 сент. 2018 г. в 23:58, Gaurav Bajaj <[email protected]>: > Hello, > > For second question above, you can have listener to listen AFTER_NODE_STOP > event and take action as per your logic (in your case changing BLT). > > Regards, > Gaurav > > > On 14-Sep-2018 9:33 PM, "eugene miretsky" <[email protected]> > wrote: > > Hi, > > What are best practices for handling split brain with persistence? > > 1) Does Zookeeper split brain resolver consider all nodes as the same > (client, memory only, persistent). Ideally, we want to shut down persistent > nodes only as last resort. > 2) If a persistent node is shut down, we need to remove it from baseline > topology. Are there events we can subscribe to? > > Cheers, > Eugene > > >
