Page MenuHomePhabricator

Bot is failing after VM migration
Open, Needs TriagePublicBUG REPORT

Description

Since T362281: Update commtech-bot VPS instance to Bookworm, the bot fails after processing so many thousands of pages. The logs suggest we're hitting the pageviews API too frequently:

2024-05-24 01:48:17  Attempt #7 to retry request to /api/rest_v1/metrics/pageviews/per-article/en.wikipedia/all-access/user/List%20of%20National%20Register%20of%20Historic%20Places%20in%20Grand%20County%2C%20Colorado/monthly/2024040100/2024043000. Server responded with no response. Waiting 10.50 seconds.
2024-05-24 01:48:28  Attempt #7 to retry request to /api/rest_v1/metrics/pageviews/per-article/en.wikipedia/all-access/user/Detroit%20Lions%20starting%20quarterbacks/monthly/2024040100/2024043000. Server responded with no response. Waiting 10.50 seconds.
2024-05-24 01:48:38  Attempt #8 to retry request to /api/rest_v1/metrics/pageviews/per-article/en.wikipedia/all-access/user/List%20of%20television%20stations%20in%20Norway/monthly/2024040100/2024043000. Server responded with no response. Waiting 12.00 seconds.
2024-05-24 01:48:50  Attempt #8 to retry request to /api/rest_v1/metrics/pageviews/per-article/en.wikipedia/all-access/user/National%20Register%20of%20Historic%20Places%20listings%20in%20Gunnison%20County%2C%20Colorado/monthly/2024040100/2024043000. Server responded with no response. Waiting 12.00 seconds.

I may have broken some of the throttling logic in https://github.com/wikimedia/popularpages/commit/275c9b72de03a734b96dee41baee1c3ac6c355a3

Event Timeline

For the month of May 2024, I'm manually restarting. But I plan to have this fixed for June's run. If working correctly, the bot should process all projects within a matter of only days.

MusikAnimal removed a subscriber: jhsoby.

(I'm assuming H219 shouldn't apply here)

I believe things are finally stable now, but next month's run will tell us for sure. I'll wait to resolve until then.