Gracefully handle invalid ghost fuzzer configurations#5380
Merged
Conversation
dylanjew
force-pushed
the
dylanj/invalid-fuzzer
branch
from
July 23, 2026 16:17
4f809a1 to
cf5f022
Compare
dylanjew
commented
Jul 23, 2026
|
|
||
| errors = result.get('insertErrors', []) | ||
| failed_count = len(errors) | ||
| insert_errors = result.get('insertErrors', []) |
Collaborator
Author
There was a problem hiding this comment.
name conflict with the package
vitaliset
approved these changes
Jul 23, 2026
When a job's fuzzer gets detached, it leaves a FuzzerJob mapping with an empty string for the fuzzer. These empty fuzzers were getting scheduled due to their weight remaining positive. When the workers picked up these ghost tasks, setup.py would fetch an invalid Fuzzer entity that lacked a blobstore_key and pass it to blobs.py in get_gcs_path, which triggered a TypeError. This change raises an InvalidFuzzerError up front instead, gracefully clearing the bad tasks without crashing.
dylanjew
force-pushed
the
dylanj/invalid-fuzzer
branch
from
July 23, 2026 16:31
cf5f022 to
f4b7c15
Compare
Collaborator
Author
|
@vitaliset Fixed the test |
dylanjew
added a commit
that referenced
this pull request
Jul 23, 2026
dylanjew
added a commit
that referenced
this pull request
Jul 23, 2026
#5380 fixed the issue where our queue was getting flooded with bad tasks but broke libfuzzer because engine fuzzers do not need a `blobstore_key`. Instead it skips the function that was causing an error for the misconfigured fuzzers I queried the Datastore and confirmed that when builtin=True, the blobstore_key is null for the engine fuzzers: `syzkaller, googlefuzztest, centipede, libfuzzer`. Additionally, the invalid fuzzer with `"Name/ID": "name=ochang_js_fuzzer_test"` is also there, which was what caused the initial errors This fixed the invalid FuzzerJob handling, but caused an issue with other libfuzzer fuzzers because libfuzzer has no blobstore key, but skips this look up that would fail because it has not blobstore key since `builtin = False` https://github.com/google/clusterfuzz/blob/05360c5aa9db354f6a9d83a105583eae7a230551/src/clusterfuzz/_internal/bot/tasks/setup.py#L606-L608
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
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.
When a job's fuzzer gets detached, it leaves a FuzzerJob mapping with an empty string for the fuzzer.
I'm not sure whether there were new tasks being scheduled or it was the existing ones blocking up the queue, but this PR focuses on clearing out the queue.
When the workers picked up these ghost tasks, setup.py would fetch an invalid Fuzzer entity that lacked a blobstore_key and pass it to blobs.py in get_gcs_path, which triggered a TypeError.
This change raises an InvalidFuzzerError up front instead, gracefully clearing the bad tasks without crashing.
This doesn't fix the issue where we're scheduling invalid fuzz tasks, but it should clear the queue by ACKing these bad tasks.