Automatically restart RelocateShard if it taking too long to complete MoveKeys
Open
- Dominant language
- C++
- Stars
- 16.7k
- Forks
- 1.6k
- Avg merge
- 1d 20h
- Merged PRs (30d)
- 126
Description
It is possible for a shard relocation to get stuck if one of the destination servers cannot complete the fetch keys. If the destination server completely dies, the algorithm will re-issue the movement to a new location, but if the destination server is just really slow the movement will never complete.
If a relocation is taking too long, we should try moving the shard to a different destination team.
Contributor guide
Assessment
This issue has not been assessed yet.