On Sun, 2016-12-04 at 13:49 +0000, Marco Calistri wrote:
I rolled back to [cfq] now but want to try the parameter vm.swappiness which as default is =60 and set it to =10:
cat /etc/sysctl.d/99-sysctl.conf
vm.swappiness=10
Rather than playing with tuning parameters, I'd suggest to proceed more systematically. Run some benchmarks to find out the "raw" performance of your disk for sequential and random IO. Do all this with the "noop" scheduler. I usually start out with a plain "dd" sequential read from the block device with 1M block size. If you have an unused partition, try sequential writes as well. Compare that to the device specs (or possibly with measurements you find on the net). This will give you an idea what to expect from the HW, and whether the HW is operating in the expected speed range. Next, run some simple benchmarks like bonnie++ on your file system. Be sure not to measure the page cache (use direct IO). Try different IO schedulers to see if it makes a difference. Finally, under your problematic load (virtualbox + packagekit), run iostat, vmstat or something similar and compare the disk throughput numbers you obtain with those from the previous steps. If the numbers (roughly) match the values to be expected from the HW, you'll either have to look for faster storage HW, or you need to distribute your IO load better. Especially if you have lots of RAM a lot of writes, you may want to look into the vm.dirty* sysctl settings. My personal rule of thumb is "set dirty_bytes to the number of bytes that the IO HW can write in ~1 second, and dirty_background_bytes to 10x the previous value". Regards Martin
Cheers,
Marco Nry隊Z)z{.r+맲rz^ˬzN(֜^ ޭ隊Z)z{.r+0Ǩ
-- Dr. Martin Wilck <mwilck@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg) -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org