[
https://issues.apache.org/jira/browse/FLINK-13848?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16961152#comment-16961152
]
Biao Liu commented on FLINK-13848:
----------------------------------
[~pnowojski],
{quote}Do I understand it correctly, that in the current code base this is not
an issue, because all of the A, B and C are executed synchronously from one
method?
{quote}
Yes, you are right, it's not an issue based on current implementation. But
it'll be an issue if we make the {{triggerCheckpoint}} asynchronously.
{quote}have you thought how to resolve the issue from you previous comment
(posted on 30/Aug/19 04:52)? With triggering next "A" only after the previous
"C" completes?
{quote}
That's a good question. I would like to talk about it, although I believe it's
an orthogonal issue. It indeed annoys me and makes the implementation more
complicated.
First of all, we have to guarantee the workflow is "A -> B -> C -> A ...",
because the checkpoint id sent to task should be monotonically increasing.
There are two kinds of triggering logic, the periodic checkpoint and savepoint.
We could avoid competition between periodic checkpoints by triggering
checkpoint manually or failing the later one. But we can't avoid competition
between savepoint and checkpoint anyway. The savepoint could be triggered
anytime and it can't be failed.
In brief, my solution is introducing a queue of trigger request. If the prior
trigger request is not finished, the later one (including checkpoint and
savepoint) will be kept in this queue. The queue will be checked when the prior
trigger request is finished. It would be included in
https://issues.apache.org/jira/browse/FLINK-13905.
> Support “scheduleAtFixedRate/scheduleAtFixedDelay” in
> RpcEndpoint#MainThreadExecutor
> ------------------------------------------------------------------------------------
>
> Key: FLINK-13848
> URL: https://issues.apache.org/jira/browse/FLINK-13848
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Coordination
> Reporter: Biao Liu
> Assignee: Biao Liu
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.10.0
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Currently the methods “scheduleAtFixedRate/scheduleAtFixedDelay" of
> {{RpcEndpoint#MainThreadExecutor}} are not implemented. Because there was no
> requirement on them before.
> Now we are planning to implement these methods to support periodic checkpoint
> triggering.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
