config:revert empty stat update fix

[ramaraochavali]

Signed-off-by: Rama rama.rao@salesforce.com
Description: Reverting the empty stat update fix so that we fix this bug #4485
Risk Level: Low
Testing: Automated tests
Docs Changes: N/A
Release Notes: N/A
Fixes #4485

[ramaraochavali]

@htuch reverted that fix

[htuch]

@ramaraochavali thanks for this. @rfaulk can you confirm this fixes the issue (if you can't patch it onto the internal tree, just make the changes by hand, it's small)?

I'd prefer not to make too many changes to how these stats are computed, as consumers will see anomalies. While @rfaulk confirms, maybe we can try and figure out the right long term fix. I would argue as a strawman that the previous "empty stat bump" could be considered correct once we have incremental xDS, since the right way to provide partial updates will be via that mechanism. Does that sound too crazy?

[ramaraochavali]

@htuch I agree that "empty stat bump" could be correct with incremental xDS. I am fine to wait on this till increments xDS comes and we can fix it if it is still confusing then.

[htuch]

@ramaraochavali OK. Let's wait for @rfaulk to confirm that this does fix the bug and we can ship the revert. Thanks.

[ramaraochavali]

Ok. No problem. I was able to reproduce the problem locally and this change fixed it. So we should be good.

[htuch]

OK, SGTM, will merge, thanks.