baa87db5d1
This commit refactors the global checkpont tracker to make it more resilient. The main idea is to make it more explicit what state is actually captured and how that state is updated through replication/cluster state updates etc. It also fixes the issue where the local checkpoint information is not being updated when a shard becomes primary. The primary relocation handoff becomes very simple too, we can just verbatim copy over the internal state. Relates #25468 |
||
|---|---|---|
| .. | ||
| licenses | ||
| src | ||
| build.gradle | ||