You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by sihuazhou <gi...@git.apache.org> on 2018/03/28 12:24:58 UTC
[GitHub] flink pull request #5781: [FLINK-9102][FLIP6] disable queued scheduling for ...
GitHub user sihuazhou opened a pull request:
https://github.com/apache/flink/pull/5781
[FLINK-9102][FLIP6] disable queued scheduling for JobGraph in Flip6LocalStreamEnvironment
## What is the purpose of the change
When we start cluster locally with fixed TMS and build stream job with Flip6LocalStreamEnvironment, we should disable queued scheduling for JobGraph.
## Brief change log
- disable queued scheduling for JobGraph in Flip6LocalStreamEnvironment
## Verifying this change
This change is a trivial rework / code cleanup without any test coverage.
## Does this pull request potentially affect one of the following parts:
- Dependencies (does it add or upgrade a dependency): (no)
- The public API, i.e., is any changed class annotated with `@Public(Evolving)`: (no)
- The serializers: (no)
- The runtime per-record code paths (performance sensitive): (no)
- Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Yarn/Mesos, ZooKeeper: (no)
- The S3 file system connector: (no)
## Documentation
- no
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/sihuazhou/flink disableQueuedSchedulingForFlip6LocalStreamEnvironment
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/5781.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #5781
----
----
---
[GitHub] flink issue #5781: [FLINK-9102][FLIP6] disable queued scheduling for JobGrap...
Posted by sihuazhou <gi...@git.apache.org>.
Github user sihuazhou commented on the issue:
https://github.com/apache/flink/pull/5781
CC: @tillrohrmann
---
[GitHub] flink issue #5781: [FLINK-9102][FLIP6] disable queued scheduling for JobGrap...
Posted by tillrohrmann <gi...@git.apache.org>.
Github user tillrohrmann commented on the issue:
https://github.com/apache/flink/pull/5781
We don't know a priori how many `TaskExecutors` have been started by the start up script. It could also be the case that some of the will be started manually at a later point in time.
---
[GitHub] flink pull request #5781: [FLINK-9102][FLIP6] disable queued scheduling for ...
Posted by sihuazhou <gi...@git.apache.org>.
Github user sihuazhou closed the pull request at:
https://github.com/apache/flink/pull/5781
---
[GitHub] flink issue #5781: [FLINK-9102][FLIP6] disable queued scheduling for JobGrap...
Posted by tillrohrmann <gi...@git.apache.org>.
Github user tillrohrmann commented on the issue:
https://github.com/apache/flink/pull/5781
This is a good question and so far I couldn't find a good solution for it. We could have different slot timeouts for standalone vs non-standalone. But this might be a bit tricky to maintain. Moreover, this would be a kind of auto magic which in general should be avoided.
---
[GitHub] flink issue #5781: [FLINK-9102][FLIP6] disable queued scheduling for JobGrap...
Posted by sihuazhou <gi...@git.apache.org>.
Github user sihuazhou commented on the issue:
https://github.com/apache/flink/pull/5781
Thanks a lot for you reply, closing this PR since it's invalid.
---
[GitHub] flink issue #5781: [FLINK-9102][FLIP6] disable queued scheduling for JobGrap...
Posted by sihuazhou <gi...@git.apache.org>.
Github user sihuazhou commented on the issue:
https://github.com/apache/flink/pull/5781
Hi @tillrohrmann you are right that seems to be impossible...but I think maybe we should at lest fail the job immediately when the job's parallelism exceeds the max slots of the fixed cluster. What do you think?
---
[GitHub] flink issue #5781: [FLINK-9102][FLIP6] disable queued scheduling for JobGrap...
Posted by sihuazhou <gi...@git.apache.org>.
Github user sihuazhou commented on the issue:
https://github.com/apache/flink/pull/5781
Hi @tillrohrmann , it's a bit sticky, do you have any idea for the current situation? That is when we submit a job and it's parallelism exceeds the max slots of the fixed cluster, we have to wait until timeout (default 5 min)... this seems not so nice to the user, and if user submit the job in detached mode they may even think that the job are running successfully
---
[GitHub] flink issue #5781: [FLINK-9102][FLIP6] disable queued scheduling for JobGrap...
Posted by tillrohrmann <gi...@git.apache.org>.
Github user tillrohrmann commented on the issue:
https://github.com/apache/flink/pull/5781
Hi @sihuazhou, I'm not sure whether this is possible, because the TMs might not be registered when you submit a job. Thus, we might need queued scheduling as well.
---