DM-34141: Distinguish retriable and non-retriable failures in Prompt Processing #100
Conversation
This file contains hidden or 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
kfindeisen
force-pushed
the
tickets/DM-34141
branch
2 times, most recently
from
7dfb148 to
59d3d8e
Compare
Originally, export_outputs would only be called if the caller knew there were datasets to export, so not finding any files implied the data ID was wrong. Now that export_outputs may be called when the pipeline failed at the beginning, this gives false positives. The original message remains as a warning so that humans can check if it came up unexpectedly.
These exceptions are intended as wrappers, to keep activator from needing to know details of what exceptions the Middleware/pipeline code can raise.
kfindeisen
force-pushed
the
tickets/DM-34141
branch
3 times, most recently
from
e13db7e to
f8db1d9
Compare
dspeck1
approved these changes
Nov 30, 2023
These handlers determine whether the error is recoverable, and ensure that any partial outputs are correctly synced to the central repo. Exceptions raised outside of pipeline execution are still the responsibility of the generic Flask handler.
This ensures that failures that have made permanent changes will never be retried.
This covers states that have modified any of the APDB, the alert stream, or the central repo.
Now that _export_subset uses Butler.transfer_from, it emits a log even if there's nothing to export. This leads to a confusing log stream. _export_subset can handle multiple input collections, so just handle them all at once.
kfindeisen
force-pushed
the
tickets/DM-34141
branch
from
f8db1d9 to
571b22c
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
This PR defines two new exceptions that let Prompt Processing specify whether a run should be retried, and adds exception handlers that understand them. The handlers are quite clumsy (in part because the desired behavior requires external I/O), so any suggestions for streamlining them would be most welcome.
This PR also cleans up our export implementation to prevent the confusing "exporting 0 datasets" messages that appeared previously.