[Bug]: CN scale out takes effect in about 10s

[dengn]

Is there an existing issue for the same bug?

• [X] I have checked the existing issues.

Environment

- Version or commit-id (e.g. v0.1.0 or 8b23a93): nightly-e033a55
- Hardware parameters:
- OS type:
- Others:

Actual Behavior

I have a cluster with 1 CN replica at first, then I set this CN replica to 2. I connect MO by proxy address, and run show backend servers to watch running CN numbers.

It appears that the CN scale out becomes effective after 10s.

This behavior will affect the speed of the cluster scaling. Every time a CN scaling needs 10 seconds+, user will experience quite a delay of 10 seconds+. We need to find a way to reduce this delay.

Expected Behavior

No response

Steps to Reproduce

No response

Additional information

No response

[volgariver6]

no plan for this issue, move to next phase plz

[dengn]

@aylei please check the current status for this issue.

[aylei]

Proxy should support watch-based CN status refresh. Currently, CN status refresh is poll-based, if a CN is labeled to a CN group, proxy can only discover the change til next poll.

[aressu1985]

s1级别的问题，delay到1.3.0在做跟踪

[volgariver6]

done