1 To choose IO schedulers at boot time, use the argument 'elevator=deadline'.
2 'noop', 'as' and 'cfq' (the default) are also available. IO schedulers are
3 assigned globally at boot time only presently.
5 Each io queue has a set of io scheduler tunables associated with it. These
6 tunables control how the io scheduler works. You can find these entries
9 /sys/block/<device>/queue/iosched
11 assuming that you have sysfs mounted on /sys. If you don't have sysfs mounted,
12 you can do so by typing:
14 # mount none /sys -t sysfs
16 As of the Linux 2.6.10 kernel, it is now possible to change the
17 IO scheduler for a given block device on the fly (thus making it possible,
18 for instance, to set the CFQ scheduler for the system default, but
19 set a specific device to use the anticipatory or noop schedulers - which
20 can improve that device's throughput).
22 To set a specific scheduler, simply do this:
24 echo SCHEDNAME > /sys/block/DEV/queue/scheduler
26 where SCHEDNAME is the name of a defined IO scheduler, and DEV is the
27 device name (hda, hdb, sga, or whatever you happen to have).
29 The list of defined schedulers can be found by simply doing
30 a "cat /sys/block/DEV/queue/scheduler" - the list of valid names
31 will be displayed, with the currently selected scheduler in brackets:
33 # cat /sys/block/hda/queue/scheduler
34 noop anticipatory deadline [cfq]
35 # echo anticipatory > /sys/block/hda/queue/scheduler
36 # cat /sys/block/hda/queue/scheduler
37 noop [anticipatory] deadline cfq