[
https://issues.apache.org/jira/browse/FLINK-24611?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17448454#comment-17448454
]
Yun Tang commented on FLINK-24611:
----------------------------------
Sorry for missing the notification as I was in the two days' on-call.
[~sewen] {{Incremental state are currently only SST files, how often are those
below 20KBytes?}}
In the general case, RocksDB should not have many 20KB size files. This could
happen if checkpoint interval is too small, thus make the flushing too
frequently. On the other hand, it might happen under case that too many states
with extremely small memory capacity, write buffer manager would flush empty
memtable in advance (see FLINK-19238).
Dropping placeholder handle would make the behavior of checkpointed state size
change. Current checkpointed state size shown in the web UI is incremental
checkpoint size, if we drop the placeholder handle, we cannot know the actual
incremental checkpoint size.
If dropping placeholder handle in changelog state-backend would make things
more easier, can we only disable the placeholder handle with changelog
state-backend, but still keep it for the normal usage of RocksDB state-backend?
Changelog state-backend would not checkpoint too often. On the other hand,
flushing memtables too often in the low-memory case could show explicit poor
performance and user should detect that then.
> Prevent JM from discarding state on checkpoint abortion
> -------------------------------------------------------
>
> Key: FLINK-24611
> URL: https://issues.apache.org/jira/browse/FLINK-24611
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Checkpointing
> Affects Versions: 1.15.0
> Reporter: Roman Khachatryan
> Assignee: Roman Khachatryan
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.15.0
>
>
> ================================================================================
> MOTIVATION
> When a checkpoint is aborted, JM discards any state that was sent to it and
> wasn't used in other checkpoints. This forces incremental state backends to
> wait for confirmation from JM before re-using this state. For changelog
> backend this is even more critical.
> One approach proposed was to make backends/TMs responsible for their state,
> until it's not shared with other TMs, i.e. until rescaling (private/shared
> state ownership track: FLINK-23342 and more).
> However, that approach is quite invasive.
> An alternative solution would be:
> 1. SharedStateRegistry remembers the latest checkpoint for each shared state
> (instead of usage count currently)
> 2. CompletedCheckpointStore notifies it about the lowest valid checkpoint (on
> subsumption)
> 3. SharedStateRegistry then discards any state associated with the lower
> (subsumed/aborted) checkpoints
> So the aborted checkpoint can only be discarded after some subsequent
> successful checkpoint (which can mark state as used).
> Mostly JM code is changed. IncrementalRemoteKeyedStateHandle.discard needs to
> be adjusted.
> Backends don't re-upload state.
> ================================================================================
> IMPLEMENTATION CONSIDERATIONS
> On subsumption, JM needs to find all the unused state and discard it.
> This can either be done by
> 1) simply traversing all entries; or by
> 2) maintaining a set of entries per checkpoint (e.g. SortedMap<Long,
> Set<K>>). This allows to skip unnecessary traversal at the cost of higher
> memory usage
> In both cases:
> - each entry stores last checkpoint ID it was used in (long)
> - key is hashed (even with plain traversal, map.entrySet.iterator.remove()
> computes hash internally)
> Because of the need to maintain secondary sets, (2) isn't asymptotically
> better than (1), and is likely worse in practice and requires more memory
> (see discussion in the comments). So approach (1) seems reasonable.
> ================================================================================
> CORNER CASES
> The following cases shouldn't pose any difficulties:
> 1. Recovery, re-scaling, and state used by not all or by no tasks - we still
> register all states on recovery even after FLINK-22483/FLINK-24086
> 2. Cross-task state sharing - not an issue as long as everything is managed
> by JM
> 3. Dependencies between SharedStateRegistry and CompletedCheckpointStore -
> simple after FLINK-24086
> 4. Multiple concurrent checkpoints (below)
> Consider the following case:
> (nr. concurrent checkpoints > 1)
> 1. checkpoint 1 starts, TM reports that it uses file f1; checkpoint 1 gets
> aborted - f1 is now tracked
> 2. savepoint 2 starts, it *will* use f1
> 3. checkpoint 3 starts and finishes; it does NOT use file f1
> When a checkpoint finishes, all pending checkpoints before it are aborted -
> but not savepoints.
> Savepoints currently are NOT incremental. And in the future, incremental
> savepoints shouldn't share any artifacts with checkpionts.
> The following should be kept in mind:
> 1. On job cancellation, state of aborted checkpoints should be cleaned up
> explicitly
> 2. Savepoints should be ignored and not change
> CheckpointStore.lowestCheckpointID
> 3. In case of JM failover, there might be more state left undeleted (see
> follow-up FLINK-24852 and/or comments)
> 4. To handle JM leadership change, backends should NOT send
> PlaceholderStreamStateHandles unless the checkpoint was confirmed (see
> comments)
> 5. To handle TM failover after an incomplete checkpoint, JM should replace
> old state handle with a new one in case of collision (test that old state
> isn't included into a completed checkpoint) (see comments)
> ================================================================================
> USER IMPACT
> For the end users, this change might render as a delay in discarding state of
> aborted checkpoints and in slight increase of undeleted state in case of
> failures; which seems acceptable.
> After updating backends to not re-upload state, checkpointing time should be
> reduced and less IO will be used (in cases when notifications are delayed or
> new JM is elected).
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
