Just wanted to throw this out there to see if anyone has any input:

I've been experiencing consistent lag on my subscriber nodes, and no  
amount of VACUUMing (full, and just analyze) has been able to make  
the lag go away.

The st_lag_num_events would slowly climb up from 1 to 10, then it  
would jump down to 1 again (sometimes it might jump down to ~5).   
Then, it would slowly work itself up to ~10 again, in an endless  
cycle.  It would very rarely hit 0, and when it did, it would only be  
for a second, then work its way back up.

What eventually caused the cyclical rising and falling was when I  
added new tables to my schema in to replication.  After creating the  
temporary set and subscribing to it, the lag would simply disappear  
after the copy_set was finished.  Now, it sticks at 0, and  
occasionally blips at 1.

I've seen this happen consistently (3 times, to be exact).  I find it  
interesting, but have no explanation for it.

Anyone have any explanations?
--Richard