Re: Group Replicaton - Lagging Slave

[Adam Nelson]

We did move from the classic master/slave replication (running on 5.5) to this setup. The old setup was on older hardware and we made the decision to move to group replication with the new hardware. The old setup did not have slave_parallel_workers setup but generally only lagged 10 to 15 seconds behind the master per the output of 'show slave status'. That being said, the old system was limited to 200 active worker connections, on the shiny new systems we bumped it up to 250, which is when this started. After making those changes we pushed it up to 400 and it's still not showing any signs of falling behind.