Almost the weekend, just a whole day to get through
@J.D. Be careful here, there is a difference between hardening and redo. Typically synchronization will mean hardening of which your answer doesn't say anything about - and no a select will not directly cause an impact to the hardening process.
Their query shows no concept of what "lag time" is, most likely it's REDO given the screenshot but no one knows
@SeanGallardy hmm, good points. I've been out of the AG game for a while now, and a dolphin's brain can only hold so much information. I understand what you're saying at a high level, but I'll probably botch it worse if I try to communicate your point. Any chance you want to edit my answer to include those points?
See, with my dolphin sized brain, I would've assumed by the word hardening, it was the other way around. Good thing I don't manage this stuff anymore 😆
Naming things is hard-ening...I guess I see what redo means in this context then. I just would've thought redo meant something along the lines of redo-ing the changes from the primary, primitively speaking, and then hardening is the final commit. But yea I'm just a dolphin.
I fixed a couple descriptions for XE items. If you all have any other XE descriptions that need fixed or improved please let me know. Unfortunately since 2019 is end of mainstream support it will only show up in 2022+.
Probably the most useful thing would be event chaining, so you could fire off one event when another event is hit. For example, I’d love to be able to collect query plans for queries when the blocked process or deadlock report is triggered so I can troubleshoot issues better and so I don’t have to source full details from several sources.
It would also be nice to collect the full query text since it’s often cut off in query plans and in the XML representation
Perhaps it would be better if Query Store collected blocking and deadlocks so you don’t have to rely on all sorts of different stuff to monitor problems