Document yaml pipeline options #30490
Conversation
|
Stopping reviewer notifications for this pull request: review requested by someone other than the bot, ceding control |
| @@ -366,6 +366,8 @@ pipeline: | |||
| config: | |||
| topic: anotherPubSubTopic | |||
| format: json | |||
| options: | |||
| streaming: true | |||
There was a problem hiding this comment.
This should be implicit due to the PubSub source. I don't know that we want to recommend setting it everywhere (though +1 to the section talking about options in the docs).
There was a problem hiding this comment.
If I remember correctly, I've had issues with Pub/Sub pipelines when the streaming option is not set. It could be an issue with xlang, but I agree it should be implicit, so it may be worth looking into fixing on the xlang side rather than explicitly defining in the YAML config.
There was a problem hiding this comment.
For python it is never implicit IIRC (we don't check whether sources are bounded or unbounded), so unless we're doing something special with yaml then I think we need this.
@tvalentyn to fact check me here, but I think we talked about making streaming implicit for Python but decided against it for back-compat reasons
There was a problem hiding this comment.
Making it implicit should not be backwards incompatible, as otherwise the pipeline wouldn't run, right?
If this is the case (we should confirm), I guess it's better to add it until the template is updated with the fix at the very least.
There was a problem hiding this comment.
@tvalentyn to fact check me here, but I think we talked about making streaming implicit for Python
I don't recall. What's the behavior in other SDKs?
There was a problem hiding this comment.
Making it implicit should not be backwards incompatible, as otherwise the pipeline wouldn't run, right?
The pipeline will just try to run in batch mode, which is actually probably fine for some unbounded sources (I think). It will run in batch mode, but generally functions fine AFAIK.
With that said, I don't really fully remember what our concerns with opting in to streaming first mode are. There are use cases here (example), though they're pretty dubious IMO.
I don't recall. What's the behavior in other SDKs?
Other SDKs opt in any unbounded sources to streaming pipelines.
There was a problem hiding this comment.
it makes sense to have a consistent behavior across SDKs. If we consider making the change, we should run some queries re: # of pipelines that had a pubsub/kafka/periodicimpulse step and were not streaming. most of those if any might be launched in error? But if there are users who intentionally run such jobs, we need to see if they will have an opt-out mechanism from new behavior.. There is a --streaming option, but no --batch option.
|
This breaks Python PreCommit again I think Python SDK relying on sources under website (#30198) is not good |
Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:
addresses #123), if applicable. This will automatically add a link to the pull request in the issue. If you would like the issue to automatically close on merging the pull request, commentfixes #<ISSUE NUMBER>instead.CHANGES.mdwith noteworthy changes.See the Contributor Guide for more tips on how to make review process smoother.
To check the build health, please visit https://github.com/apache/beam/blob/master/.test-infra/BUILD_STATUS.md
GitHub Actions Tests Status (on master branch)
See CI.md for more information about GitHub Actions CI or the workflows README to see a list of phrases to trigger workflows.