numinnex commented on PR #3425: URL: https://github.com/apache/iggy/pull/3425#issuecomment-4668625083
Ok, the idea is sound and looks good, there is one piece that I think is worth addressing at this level -- `replica_id` generation. Currently we are supplying the replica_id through an CLI argument, it's fine for an "seed" cluster values (e.g perfect information first bootstrap), but falls flat in cases where we would like to have dynamic cluster counts. One way to address this isssue would be to keep the initial seed, but on the first bootstrap (when the metadata log is empty, so the cluster is fresh), write that seed configuration into the metadata log with op=0. Make the `replica_id` be generated at runtime, rather than supplied (the one comming from CLI arg would be used as an index into the `replicas` set, so the initial seed configuration has granular control over all of the nodes -- required for determinism). After the initial configuration is written to the log and resolved, we can go through the `challenge` protocol with those nodes). I omitted one important detail which is `ReplicaJoin` command, as this is something that should be addressed with other PR. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
