dawidwys commented on a change in pull request #18086:
URL: https://github.com/apache/flink/pull/18086#discussion_r768698948
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/minicluster/MiniCluster.java
##########
@@ -875,6 +890,21 @@ public JobExecutionResult executeJobBlocking(JobGraph job)
(Acknowledge ignored) -> new
JobSubmissionResult(jobGraph.getJobID()));
}
+ // HACK: temporary hack to make the randomized changelog state backend
tests work with forced
+ // full snapshots. This option should be removed once changelog state
backend supports forced
+ // full snapshots
+ private void checkRestoreModeForRandomizedChangelogStateBackend(JobGraph
jobGraph) {
+ final SavepointRestoreSettings savepointRestoreSettings =
+ jobGraph.getSavepointRestoreSettings();
+ if (overrideRestoreModeForRandomizedChangelogStateBackend
+ && savepointRestoreSettings.getRestoreMode() ==
RestoreMode.NO_CLAIM) {
+ final Configuration conf = new Configuration();
+ SavepointRestoreSettings.toConfiguration(savepointRestoreSettings,
conf);
+ conf.set(SavepointConfigOptions.RESTORE_MODE, RestoreMode.LEGACY);
Review comment:
It has been discussed. The reason why we do want to go for the
`NO_CLAIM` mode by default is that the current `LEGACY` is inherently broken
and really misleading and unexpected. You can basically never safely delete an
externalized incremental checkpoint.
--
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: issues-unsubscr...@flink.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
