MySQL Forums
Forum List  »  Replication

Re: 5.6 to 5.7 row-based replication performance issue
Posted by: JOSEPH MASK
Date: October 28, 2020 04:45PM

> Do you have a replication test setup where you can benchmark a 5.7 primary replicating to a 5.7 replica?

Yes, I set up a 5.7 master and I see the same problem with performance of a 5.7 replica compared to that of a 5.6 replica. It took the 5.7 replica 73 minutes to replicate 60 minutes of row-based binlog from the 5.7 master, but it only took the 5.6 replica 15 minutes to do the same.

> See if doing pager grep Query, then show processlist; identifies queries that are slow, and/or enable performance_schama SQL,IO,Replication instruments.

On closer observation of show processlist, system lock is happening with about the same frequency on both the 5.6 and 5.7 replicas. The same slow queries show up on both replicas also. If I remember right, we have performance_schema disabled because it causes mysqld startup to take an hour or so, since there are so many tables it has to scan. As a result, we're not too familiar with its usage, or the SQL,IO,Replication instruments you referenced, but I'll read up on them and see if I can make use of them. Thanks for the pointers so far, and feel free to throw out any more ideas you may have.

Options: ReplyQuote




Sorry, you can't reply to this topic. It has been closed.

Content reproduced on this site is the property of the respective copyright holders. It is not reviewed in advance by Oracle and does not necessarily represent the opinion of Oracle or any other party.