| Arthur Bello | 27c5175 | 2023-09-06 18:14:19 | [diff] [blame] | 1 | # CQ Fault Attribution |
| 2 | |
| 3 | The Failure Analysis section in Gerrit provides an easily discoverable view of |
| 4 | CQ test variant (and soon, build) failures, and categorizes them by comparing |
| 5 | the failures to the same test variant in snapshot builds. The failures are |
| 6 | categorized as one of the following: |
| 7 | |
| 8 | - New failure (not currently present in tip-of-tree \[ToT\]) |
| 9 | - Pre-existing with a different failure reason (test variant failure present in |
| 10 | ToT, but for a different failure reason) |
| 11 | - Pre-existing with the same failure reason (test variant failure present in |
| 12 | ToT for the same failure reason) |
| 13 | |
| 14 | Additionally, any test variant failures that have not been able to be |
| 15 | classified due to insufficient test history in the snapshot / postsubmit builds |
| 16 | used in ToT will be indicated as such. |
| 17 | |
| 18 | Within the Gerrit checks tab, any fault attributed test row will contain a tag |
| 19 | with the corresponding fault attribution category. For any tests pre-existing |
| 20 | failures, a link to the snapshot where the failure was found will be present as |
| 21 | well. |
| 22 | |
| 23 | To provide any feedback on fault attribution (either positive or constructive |
| 24 | feedback), please feel free to create a buganizer issue [here](https://buganizer.corp.google.com/issues/new?component=1315651&template=1849636). |
