We recently upgraded to b5845 from b5794.

This was performed on 2 clusters (4 machines) and immediately on both, saw increased IO wait times.

After playing around with everything and anything possible, and ruling out hardware failure as this is happening on 4 separate machines, even with failover turned off to see if was an issue with the replication, we downgraded.

On downgrade, IO went back to normal and things were good.

Are there any known issues with b5845 relating to IO performance?