Currently running a SQL 2014 instance Availability Group(s) with a single synchronous replica.
We're having some issues, and I"m seeing a large number of HADR_SYNC_COMMIT waits. In an attempt to troubleshoot things it appears that there may be a latency issue so I'm trying to disable the Log Compression to see if there are any changes. I know that in SQL 2016 this is disabled by default for Synchronous replicas. However in 2014 and earlier it was enabled by default.
So i've set the trace flag via
dbcc traceon (1462, -1)
I then run some AlwaysOn Latency Data Collection extended events on the primary to check and see if this worked. Upon reviewing the extended events I still see hadr_log_block_compression. https://blogs.msdn.microsoft.com/sql_server_team/troubleshooting-high-hadr_sync_commit-wait-type-with-always-on-availability-groups/
Occurs when log is ready to be captured and log compression is enabled. (If log compression is not enabled, this XEvent is not logged in SQL12 and SQL14, but it is always logged from SQL16. For SQL16, check Boolean value of property is_compressed.)
I still see these events and according to the above link I shouldn't even see them. However there is a field is_compressed for that event that is has a value ofFalse.
What's going on here? Does the flag not function for Synchronous replicas on sql 2014, or is it working and is the documentation actually incorrect?