Hello, welcome to another update. Last week I managed to finish updating the Cabal release pipelines, and I’m continuing to categorize spurious failures on GHC CI.
Most recently, I looked at cases where jobs fail because GitLab responds with HTTP error codes. My first priority is to track them and automatically retry failed jobs, but eventually I will want to investigate the root cause and fix it. This class of failures, and the one where jobs are unexpectedly killed with signal 9, are the first two that I want to investigate once I am finished categorizing all reported spurious failures. Speaking of which, I have currently investigated half of all reports (84 of 168), and it even seems that I may be cleaning them up faster than new ones are coming in.
Wish me luck!