NSAmelchev opened a new pull request, #13808: URL: https://github.com/apache/ignite/pull/13808
`SnapshotCompatibilityTest#testSnapshotRestore` is flaky: a snapshot restore started right after the cluster activation fails with "Cache group restore operation was rejected. The cluster should be active." `IgniteCluster#state` returns when all nodes have finished the state change exchange, but the local cluster state stays in transition until `ChangeGlobalStateFinishMessage` is processed. `SnapshotRestoreProcess#start` checked the transition flag in the caller's thread and rejected the operation. The transition check is removed from `SnapshotRestoreProcess#start`. The cluster state is validated again on the prepare phase, which is ordered with the finish message by discovery, so the restore is still rejected if the state change is not finished by then. Behavior change: in that case the restore is rejected with the same message on the prepare phase, i.e. after the snapshot check, instead of immediately. The workaround for the same race is removed from `IgniteClusterSnapshotRestoreSelfTest#testUserClassRestored`. -- 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]
