Github user StefanRRichter commented on a diff in the pull request:
https://github.com/apache/flink/pull/3531#discussion_r107873583
--- Diff:
flink-runtime/src/main/java/org/apache/flink/runtime/state/heap/HeapKeyedStateBackend.java
---
@@ -412,9 +421,15 @@ private void
restorePartitionedState(Collection<KeyGroupsStateHandle> state) thr
}
}
- for (Tuple2<Integer, Long> groupOffset :
keyGroupsHandle.getGroupRangeOffsets()) {
+ for (Tuple2<Integer, Long> groupOffset :
keyGroupsStateHandle.getGroupRangeOffsets()) {
int keyGroupIndex = groupOffset.f0;
long offset = groupOffset.f1;
+
+ // Skip those key groups that don't
belong to the backend.
+ if
(!keyGroupRange.contains(keyGroupIndex)) {
--- End diff --
I think this is no longer required, because we are now "cutting out" the
right key-groups from each state handle in the `StateAssignmentOperation`,
using the `KeyedStateHandle::getIntersection(...)` method. In fact, I think
that receiving a key-group that is not in the backend's range could now be
considered as an error. We could rewrite this as a precondition check, or an
assert.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
