Re: sanity check on innodb config

[kevin foote]

So .. not sure if innodb_fast_shutdown is what I really want to do then, but glad to know its there.

I think O_DIRECT will help with the minimal swaping that I'm seeing. Using vmstat si and so are only in the range of 20 to 50, not 1k to 2k like the 'High Performance MySQL' book depicts. I might also throw in setting swappiness to 0 rather than the default 60..


One other observation on my innodb / mysql server set up.

While innotop reports mysql as having ~ 4.5g allocated via the "Buffer Pool" view page, the OS is reporting RES (Resident Memory) usage never going above ~ 2.4g

--- top output ---
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
25892 mysql 15 0 4996m 2.2g 18m S 5.7 29.1 67:35.80 mysqld

Am I hitting a restrictive limit via the OS? (previously mentioned I have not messed with sysctl etc) Is this due to paging size or some sysctl setting that I'm missing? Is this even relevant on x86_64?

>
> Why do you need to stop mysql more than, say, once a year?
>

Well once I get all this ironed out I don't really see the need but currently I do need to restart things. Which gives the pain of 10 or 20 minutes..

One other reason to restart that I've had lately is to disconnect clients. Haven't found a way to do that within mysql yet.. :-) FLUSH HOSTS just flushes the host knowledge I think.. errors etc..


Thanks for all the help on this..
kevin