Rework pipeline to use offline triaging
- assemble checkout_all structure for one individual build and its tests
- create appropriate issueoccurences by checking issueregexes
- save the kcidb_all.json data
- hand this to reporter to create the report
AC:
- one check-results job per arch, no wait-for-dw
- triager can get DW object to triage do load-from-file or load-from-DW
- with a fully triaged kcidb_all.json file in the artifacts that contains the issue occurances
- no wait-for-triaging
- MR comments are updated whenever a check-results job finishes, based on the data contained in the kcidb_all.json file in there
Proposed pipeline structure:
A single stage for triaging + reporting, one job per architecture
Communication between triager and DW via file for CVE, via DW otherwise
Proposed data paths:
- CVE:
- Triager triages from file, specified build ID
- Triager triages to file
- Reporter reports from file, [specified build ID]
- Otherwise:
- Triager triages from file, specified build ID
- Triager triages to file and tries to submit to DW
- Reporter reports from DW with a specified build ID
Things that still need implementing
-
Reporter `--build-id` option reporter!132 (merged) -
Add aliases to triager CLI options (drop "offline"/"online" mentions because they are confusing) -
Rework !1899 into single stage (triage+report), one job per arch again
Edited by Ondřej Kinšt