1554136 - Error saving classification shows up when classifying fails

Status: RESOLVED FIXED

(Tree Management :: Treeherder, defect, P2)

Description

[Cosmin Sabou [:CosminS]]

Attachment: image.png

this doesn't happen all the time but mostly when classifying failures quickly one after another, please see the attached screenshot.
also noticed that this appears when you have multiple fails pinned and you classify them all at once.

Comment 1

[Andreea Pavel [:apavel]]

Hi Cameron, are there any updates here?

Comment 2

[Cameron Dawson [:camd]]

I haven't yet been able to reproduce this. I'll keep trying on stage. I don't suppose this has changed since the deploy 6 days ago? We pushed some changes I made in this area.

Comment 3

[Cameron Dawson [:camd]]

OK, I spoke too soon. I was finally able to reproduce this. I have not yet figured out the error though. Trying to track that down...

Comment 4

[Cameron Dawson [:camd]]

I found the error and have a fix. In reality, there was no error classifying. It was an error in trying to reapply filters AFTER classifying. But if you classify several jobs in the same platform, the later jobs will already be filtered away. But I wasn't checking if the job was gone before trying to call filter on it. So a simple check fixes it.

Comment 5

[GitHub Bugzilla PR Linker]

Attachment: Link to GitHub pull-request: https://github.com/mozilla/treeherder/pull/5085

Comment 6

[Cameron Dawson [:camd]]

I tested on stage and it looks fixed now.

Comment 7

[Andreea Pavel [:apavel]]

Hi Cameron, I could still reproduce this again this shift, same situation, when classifying multiple failures at once: https://irccloud.mozilla.com/file/tp28916N/classification_error

Comment 8

[Cameron Dawson [:camd]]

Andrea-- Sorry, I should have clarified. This change has been pushed to stage, but not production yet. I will do a production push today to get all these fixes out. If you wanted to test on treeherder.allizom.org, it should work correctly there.

Comment 9

[Andreea Pavel [:apavel]]

Ahh, okay. Tested on staging and indeed the issue is fixed. Thank you.