canKillProcesses in simulator can return a false result
Open
- Dominant language
- C++
- Stars
- 16.7k
- Forks
- 1.6k
- Avg merge
- 1d 20h
- Merged PRs (30d)
- 126
Description
We saw a failure in `SwizzledRollbackSideband.txt` (seed `391271596` and git hash `ad0b59ac7` in release-6.2) because `canKillProcesses` was returning `true` even though the processes `RemoveServersSafely` wanted to exclude and kill caused FDB to not recover anymore. Specifically, this resulted in the workload excluding enough processes that no two data halls had two machines left (which is a requirement in order to recruit tlogs in the three data hall mode).
Contributor guide
Assessment
This issue has not been assessed yet.