Open cholcombe973 opened 7 years ago
Track the latency of disks and kick slow ones out of the cluster. This can kill a cluster if disks are slow but not dead. Benchmark the disks and keep a log somewhere. If 5 times in a row of bad results, kick the disk out of the cluster
https://github.com/Comcast/Bynar/pull/77#issuecomment-515974986
Track the latency of disks and kick slow ones out of the cluster. This can kill a cluster if disks are slow but not dead. Benchmark the disks and keep a log somewhere. If 5 times in a row of bad results, kick the disk out of the cluster