Consider also servicing your requests in LIFO order, or using a very short queue. It would have prevented overload from causing complete lockup; rather you would have more gracefully shed the excess load you couldn’t handle.
After noticing a worrying pattern in the performance of Gitaly, a Git RPC service we're rolling out across http://GitLab.com , we uncovered an issue with fork locking affecting virtual memory size. Here's how we fixed it, via @suprememoocow 
https://about.gitlab.com/2018/01/23/how-a-fix-in-go-19-sped-up-our-gitaly-service-by-30x/ …
-
-
-
We've implemented a similar scheme, but currently it's per-method-per-repository, not global. Allows us to limit to `n` concurrent requests per repository and LIFO queue the rest. https://gitlab.com/gitlab-org/gitaly/merge_requests/376 …
- Pokaż odpowiedzi
Nowa rozmowa -
Wydaje się, że ładowanie zajmuje dużo czasu.
Twitter jest przeciążony lub wystąpił chwilowy problem. Spróbuj ponownie lub sprawdź status Twittera, aby uzyskać więcej informacji.