[ 
https://issues.apache.org/jira/browse/FLINK-20103?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Roman Khachatryan updated FLINK-20103:
--------------------------------------
    Description: 
This is a follow-up ticket after FLINK-20097.

The current setup (UnalignedITCase) doesn't reveal:
 - race conditions (1 per tens of runs)
 - or bugs triggered in some specific setup (low checkpoint timeout)

Changes that might hide the bugs:
 - having Preconditions in ChannelStatePersister (slow down processing)
 - some Preconditions may mask errors by causing job restart
 - timings in tests (UnalignedITCase)

An additional issue with the current setup is that it's difficult to git bisect 
(for long ranges).

  was:
This is a follow-up ticket after FLINK-20097.

The current setup (UnalignedITCase) doesn't reveal:
- race conditions (1 per tens of runs)
- or bugs triggered in some specific setup (low checkpoint timeout)

Changes that might hide the bugs:
- having Preconditions in ChannelStatePersister (slow down processing)
- some Preconditions may mask errors by causing job restart
- timings in tests (UnalignedITCase)



> Improve test coverage for network stack
> ---------------------------------------
>
>                 Key: FLINK-20103
>                 URL: https://issues.apache.org/jira/browse/FLINK-20103
>             Project: Flink
>          Issue Type: Improvement
>          Components: Runtime / Checkpointing, Runtime / Network, Tests
>            Reporter: Roman Khachatryan
>            Priority: Major
>             Fix For: 1.13.0
>
>
> This is a follow-up ticket after FLINK-20097.
> The current setup (UnalignedITCase) doesn't reveal:
>  - race conditions (1 per tens of runs)
>  - or bugs triggered in some specific setup (low checkpoint timeout)
> Changes that might hide the bugs:
>  - having Preconditions in ChannelStatePersister (slow down processing)
>  - some Preconditions may mask errors by causing job restart
>  - timings in tests (UnalignedITCase)
> An additional issue with the current setup is that it's difficult to git 
> bisect (for long ranges).



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to