Auto-drain #6
Auto-drain #6
Conversation
* master: fix drain test do not check tempfile on head fix docker setup fix fid path parsing add rabbitmq addon check tempfile on patch
|
@BatuAksoy made some changes after your review. @muraty PR is ready now. |
| func (s *Server) autoDrain() { | ||
| defer close(s.autoDrainStopped) | ||
|
|
||
| // Drainer is written for "efes drain" command but can be reused here because we need same functionality here. |
There was a problem hiding this comment.
I guess we won't need this comment anymore.
| @@ -204,34 +202,6 @@ func main() { | |||
| return nil | |||
| }, | |||
| }, | |||
| { | |||
| Name: "drain", | |||
There was a problem hiding this comment.
Since we delete this, we won't need Run function of Drainer struct anymore, right?
| // Auto-drain period is a window that only certain number of servers are allowed to run auto-drain operations. | ||
| // When the period advances to the next one, servers currently running auto-drain will stop and some other | ||
| // servers are going to run auto-drain. | ||
| AutoDrainRunPeriod Duration `toml:"auto_drain_run_period"` |
There was a problem hiding this comment.
@cenkalti how to find appropriate value for this config?
Let's say;
- We have 100 servers
- 50 of them exceeds
AutoDrainThreshold AutoDrainDeviceRatiois set to10
In this case, if we set AutoDrainRunPeriod as 1 hour, 5 servers will be drained every 1 hour. So, auto-drain will be finished in 10 hours for 50 servers. Is the calculation correct?
If there is small difference for the servers those exceeds AutoDrainThreshold and those don't, 1 hour might be a high value. Maybe we should use less than that? Or should we find appropriate AutoDrainRunPeriod config with run & monitor method?
No description provided.