I have many CentOS systems, some of them have traditional hard disks, some SSD, and some mix of them. I would like to change scheduler for SSD to deadline.
So at runtime I can simply scan check devices and for ones that have /sys/block/sd?/queue/rotational
as 0
. I could enable deadline by doing echo 'deadline' > /sys/block/sda/queue/scheduler
But what about permanent changes? Some tuned profiles enable deadline scheduler, but if there is mixed (some normal, some SSD), for which disks would this be applied?
For example:
tuned-adm profile latency-performance
latency-performance A server profile for typical latency performance tuning. This profile disables dynamic tuning mechanisms and transparent hugepages. It uses the performance governer for p-states through cpuspeed, and sets the I/O scheduler to deadline. Additionally, in Red Hat Enterprise Linux 6.5 and later, the profile requests a cpu_dma_latency value of 1. In Red Hat Enterprise Linux 6.4 and earlier, cpu_dma_latency requested a value of 0.
So if I have several disks, for which deadline would be applied?
While you could write a customer profile or modify one of the existing, I'd recommend
tuned-adm profile enterprise-storage
based on your hardware information.If you need more changes beyond that, just use "enterprise-storage" as a baseline.