Slow disk read Galera

[sixart]

Hi, I have a problem with cluster galera.

I have three node, but the disk read is very slow.
I have storage NetApp where I stored data.
My Configuration is:

• For all server I have 8 CPU and 12GB of RAM.
• My my.cnf is in attachment

How to configure it for best performance? And for more speed in disk read?

Regards,
Dario

my_cnf.txt (5.08 KB)

[przemek]

What do you mean bye “disk read is very slow”?
In general, Galera replication does not affect reads, it’s rather about writes that are affected by the network latency on commit stage.
NetApp sounds like a NFS share maybe? I don’t think NFS based disk storage is the best solution for MySQL. Still where is the problem? Did you check the bottlenecks with iostat, vmstat, mpstat or other tools maybe?

Also another general rule is that having large enough Buffer Pool to keep the most hot data may allow to avoid most of the disk reads. I can see you have:
innodb_buffer_pool_size=3990M
while 12G total RAM. Check whether the Buffer Pool is fully populated, and if there is still free RAM in the machine. Increase the BP accordingly, but do not exceed the ~80% of the total RAM. Add more RAM if necessary.