0

I have a Master-Master cluster with 2 Percona 5.5.30 servers. (Lets name them 1 & 2) The OS is FreeBSD 9.1

I have an issue that server 2 is not processing the relay log fast enough. I know that for sure, because when i create a new table on server 1, and i immidiatly check the relay-log on server 2, i see the CREATE command there, but the table is not being created for another 5-15 minutes or so.

It looks like the thread that is supposed to executing the relay log is halting itself every once in a while, showing NULL in its state when executing SHOW PROCESSLIST;

What can be the reason for that?

UPDATE

Processlist: http://pastebin.com/5Pq7f0GF

(I don't know how to paste SQL here...)

2
  • The next time this happens, please post the output of SHOW PROCESSLIST;. If the processlist is too long, then run SELECT * FROM information_schema.processlist WHERE user='system user'\G Jun 13, 2013 at 10:29
  • Added to post..
    – shaharmor
    Jun 13, 2013 at 15:40

1 Answer 1

1

This is quite common. A slave only has a single IO thread (the slave I/o thread) whereas a master is typically executing multiple queries in parallel. You should look at your disk I/O usage (like with iostat) to determine if you're bottle necked there or at the CPU (single thread can use only one core).

The 5.6 series of MySQL can use a separate thread for each schema, which may help.

1
  • So there is nothing i can do right now?
    – shaharmor
    Jun 13, 2013 at 19:33

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .