Fix current issues that cause Milo to regularly respond with 500 on user-facing endpoints |
||
Issue descriptionYesterday LuciMiloHTML5xxRateHigh fired, and it appears if it was enabled as a page alert now it might fire regularly for several routes on Milo. Before making this alert a paging alert, we should resolve current issues. "Resolving issues" may in some cases mean returning a 4xx response instead of 5xx. Some such endpoints may include: /b/:id /p/:project/builders/:bucket/:builder/:numberOrId /p/:project/g/:group/console /swarming/task/:id/steps/*logname The alert yesterday: https://groups.google.com/a/google.com/forum/#!topic/chops-foundation-alerts/cYGbdRi10nQ
,
Aug 22
i've silenced the alert because it produces too much spam. Please unsilence or adjust the alert if you disagree. http://silence/2749219513504366592 Here is the graph of human-visible HTTP 500s http://shortn/_pcmZqsi7sF
,
Aug 22
500s should be rare and indicate real problems. I looked through pantheon and see a bunch of "console not found" errors for the /p/:project/g/:group/console URLs that aren't being reported as 404s, as well as a gitiles timeout and a request timeout. It would be good to clean these up so we can see when something is broken. |
||
►
Sign in to add a comment |
||
Comment 1 by efoo@chromium.org
, Aug 21