jtuglu1 commented on PR #18819:
URL: https://github.com/apache/druid/pull/18819#issuecomment-3656761038

   > At most, maybe keep just one config scaleActionPeriod that can be 
specified as an ISO period (e.g. PT1M) or something (mostly since you would be 
using this in embedded tests). The other configs don't really add any value. 
They are legacy configs in lag-based auto-scaler which we might as well avoid 
adding in the new strategy.
   
   I don't fully agree with this. At the very least, we use 
`config.getScaleActionStartDelayMillis()` internally when doing red/black 
deployments where supervisors can get paused. It's better in our case to put a 
delay after resubmitting the supervisor, otherwise we end-up over-scaling after 
a deployment. Similarly, we update the specs frequently to add new columns. 
Putting a cooldown after submission allows the scaler to adjust accurately to 
the lag rather than getting in a scaling loop and becoming way over-scaled. I 
agree the rest are not too useful in practice.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to