chore: change replica connect error log to WARN

Open romange opened this issue 1 year ago • 3 comments

Context: reduce noise when performing H/A datastore update because we get spurious errors. We identify the disconnect event with metrics (by looking at master lag).

Jun 25 '24 06:06 romange

@romange how would you track multiple disconnect events on replication? if you have some kind of a bug in replication that causes replica to reconnect again and again. we have 0 lag in full sync mode

Jun 25 '24 07:06 adiholden

True. In that case I would like the team to take over this PR and add the metric. Also maybe fix https://github.com/dragonflydb/dragonfly/issues/3210 while we are at it.

Jun 25 '24 07:06 romange

https://github.com/dragonflydb/dragonfly/issues/3215

Jun 25 '24 08:06 adiholden