[
https://issues.apache.org/jira/browse/FLINK-9352?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16490187#comment-16490187
]
vinoyang commented on FLINK-9352:
---------------------------------
[~StephanEwen] delay random is a way to handle this issue. We provide a
streaming API to let the user control the initial time, if not setting, use the
default checkpoint interval. what do you think about this way?
> In Standalone checkpoint recover mode many jobs with same checkpoint interval
> cause IO pressure
> -----------------------------------------------------------------------------------------------
>
> Key: FLINK-9352
> URL: https://issues.apache.org/jira/browse/FLINK-9352
> Project: Flink
> Issue Type: Bug
> Components: State Backends, Checkpointing
> Reporter: vinoyang
> Assignee: vinoyang
> Priority: Critical
>
> currently, the periodic checkpoint coordinator startCheckpointScheduler uses
> *baseInterval* as the initialDelay parameter. the *baseInterval* is also the
> checkpoint interval.
> In standalone checkpoint mode, many jobs config the same checkpoint interval.
> When all jobs being recovered (the cluster restart or jobmanager leadership
> switched), all jobs' checkpoint period will tend to accordance. All jobs'
> CheckpointCoordinator would start and trigger in a approximate time point.
> This caused the high IO cost in the same time period in our production
> scenario.
> I suggest let the scheduleAtFixedRate's initial delay parameter as a API
> config which can let user scatter checkpoint in this scenario.
>
> cc [~StephanEwen] [~Zentol]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
