MySQL Chained Replication - Do Not Forget log-slave-updates
I was setting up a chained replication topology for MySQL today: Master(1) <- Master/Slave(2) <- Slave(3). After 2 hours of not seeing any changes propagate to (3) after changes on (1) and (2), I came to the conclusion that additions were not being written to (2)'s log file. SHOW SLAVE STATUS on (3) was reporting that (3) was connecting to (2) and that both the I/O and SQL threads were running. Yet no updates were happening on (3)!
After much Googling [What does one type in to find this stuff?!] and some good ol' Stack Overflow, I eventually tracked down the problem in (2)'s my.cnf. I was not logging the slave updates. I consider this a gotcha because the documentation is a bit thin on chained replication. What little I have gleaned is from various blog posts scattered from out of date web forums.
So...make sure you include log-slave-updates as a line in your my.cnf for your middle master/slave. I placed it directly below the server-id so that I knew not to forget it was there.
Have been searching cluelessly for this since 10 hours or so. This post really saved half-my-remaining-day :)
Same here; I was searching for a solution for several hours. Found something about those log slave updates on old webblogs, but your post is the first which is clearly.
Replication is working flawless now :-) .
Thanks a lot!
Oh and for god sake make sure you get the correct option in the my.cnf
as I experienced the "ON" and "1" are different between versions of mysql
5.5
log-slave-updates = ON
5.1
log-slave-updates = 1
it does produce an error but I missed it scrolling past on a producetion server :(