Re: Mysql use up all my swap space WHY?

[yvonne kire]

Thanks for you input
Appreciate it :)

What's been done since yesterday:
mysql> RESET QUERY CACHE;
mysql> FLUSH QUERY CACHE;
echo 0 > /proc/sys/vm/swappiness (was 1)

That gain back some SWAP space enough to clear the TIVOLI alert
but we still have too much SWAP space being used.....


PLAN-A
I will bump up the innodb_buffer_pool_size a bit
before I restart the mysqld

PLAN-B
I see the other slaves in the cluster has much more RAM
and that's probably the reason why they keep up and doesn't fall behind the Master ie rule : have more powerfull slaves than the master or you will see replication lag on a busy cluster...

I could promote slave #2 to a new master with 98GB RAM and I probably have to retire the old Master with 48GB that will probably not hold up as slave.


PLAN-C
Build a new replacement master with much more RAM (if we still see swap issues)

Here is the Cluster host setup with #1 being the master
I also see the master is a FusionIO host (R610f) with RAM is 48GB.



Num Server_ID RAM_Size HD_Size HD_Used HD_Free Used% Speed Cache HT MySQL OSbits Linux_Version Model
1 ki307-22 49408748 689G 307G 348G 47% 2.40GHz 12288 16 5.5.32 x86_64 CentOS6.9(Final) R610f
2 ki111-36 99051240 720G 292G 392G 43% 2.40GHz 12288 16 5.5.32 x86_64 CentOS6.9(Final) R710
3 ki213-37 148691620 720G 292G 392G 43% 2.40GHz 12288 16 5.5.32 x86_64 CentOS6.9(Final) R710
4 ki302-37 148691620 689G 306G 349G 47% 2.40GHz 12288 16 5.5.32 x86_64 CentOS6.9(Final) R710f
5 ki303-37 99051208 720G 306G 378G 45% 2.40GHz 12288 16 5.5.32 x86_64 CentOS6.9(Final) R710