[
https://issues.apache.org/jira/browse/FLINK-27681?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17755766#comment-17755766
]
Hangxiang Yu commented on FLINK-27681:
--------------------------------------
Hi, [~Ming Li] Thanks for reporting this.
We also saw some similar cases.
I think the core reason is lacking checking checksum when checkpointing , so
even if the checksum is incorrect, error files are still be uploaded but
checkpoints still be marked as completed.
{quote}Can RocksDB periodically check whether all files are correct and find
the problem in time?
{quote}
I think It's better to check every CP files that will be uploaded so that we
could avoid local corrupted files affecting remote checkpoint files. But this
may increase costs and decrease the performace of checkpoints. So maybe a
configuration could be introduced ? WDYT?
{quote}Can Flink automatically roll back to the previous checkpoint when there
is a problem with the checkpoint data, because even with manual intervention,
it just tries to recover from the existing checkpoint or discard the entire
state.
{quote}
I think it makes sense, but just as I said, we must make sure that every
checkpoints will not be influenced by the local corrupted files.
> Improve the availability of Flink when the RocksDB file is corrupted.
> ---------------------------------------------------------------------
>
> Key: FLINK-27681
> URL: https://issues.apache.org/jira/browse/FLINK-27681
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / State Backends
> Reporter: Ming Li
> Priority: Critical
>
> We have encountered several times when the RocksDB checksum does not match or
> the block verification fails when the job is restored. The reason for this
> situation is generally that there are some problems with the machine where
> the task is located, which causes the files uploaded to HDFS to be incorrect,
> but it has been a long time (a dozen minutes to half an hour) when we found
> this problem. I'm not sure if anyone else has had a similar problem.
> Since this file is referenced by incremental checkpoints for a long time,
> when the maximum number of checkpoints reserved is exceeded, we can only use
> this file until it is no longer referenced. When the job failed, it cannot be
> recovered.
> Therefore we consider:
> 1. Can RocksDB periodically check whether all files are correct and find the
> problem in time?
> 2. Can Flink automatically roll back to the previous checkpoint when there is
> a problem with the checkpoint data, because even with manual intervention, it
> just tries to recover from the existing checkpoint or discard the entire
> state.
> 3. Can we increase the maximum number of references to a file based on the
> maximum number of checkpoints reserved? When the number of references exceeds
> the maximum number of checkpoints -1, the Task side is required to upload a
> new file for this reference. Not sure if this way will ensure that the new
> file we upload will be correct.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
