This repository has been archived by the owner on Oct 9, 2023. It is now read-only.
Abort subworkflow on subnode failure #468
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Daniel Rammer <[email protected]>
…e in ui Signed-off-by: Daniel Rammer <[email protected]>
Signed-off-by: Daniel Rammer <[email protected]>
Signed-off-by: Daniel Rammer <[email protected]>
Signed-off-by: Daniel Rammer <[email protected]>
2 tasks
|
Ohh man this is a very good catch |
|
This is a very good finding! Unfortunately I do not have sufficient knowledge to do a constructive review, but I'm looking forward to testing it after it is merged. |
|
Hey @hamersaw I confirmed by checking some stuck Pods in our cluster currently. |
Signed-off-by: Daniel Rammer <[email protected]>
|
Shall we get this going? It would be great to get this fix so we don't need to have a hacky way to clean up terminating pods. Thanks. |
EngHabu
approved these changes
Sep 7, 2022
eapolinario
pushed a commit
to eapolinario/flytepropeller
that referenced
this pull request
Aug 9, 2023
* using 'failing' state to handle subworkflow aborts Signed-off-by: Daniel Rammer <[email protected]> * propogating node failure is subworkflow to subworkflow failure message in ui Signed-off-by: Daniel Rammer <[email protected]> * working with other failure scenarios Signed-off-by: Daniel Rammer <[email protected]> * fixed lint issue Signed-off-by: Daniel Rammer <[email protected]> * updated error message to match Signed-off-by: Daniel Rammer <[email protected]> Signed-off-by: Daniel Rammer <[email protected]>
Closed
2 tasks
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
TL;DR
Currently, if an internal node in a subworkflow fails the parent workflow aborts. However, this abort does not traverse into the subworkflow to abort other nodes that may be running. This means that Flyte Pods may be left orphaned, this is especially important if finalizers are set and Flyte never cleans up the Pods.
Type
Are all requirements met?
Complete description
We updated the failure handling in the subworkflow to match the current dynamic task implementation. If an internal node (in the subworkflow) fails we mark the WorkflowNodeState as failing but keep the NodeState as running. Then in the next iteration FlytePropeller processes the failing subworkflow by first aborting it and then processing the failure node (if it ever exists). Errors are propagated up the workflow identically in failing subworkflows and dynamic tasks.
Tracking Issue
fixes flyteorg/flyte#2533
fixes flyteorg/flyte#2574
Follow-up issue
NA