7 cgroup subsys "blkio" implements the block io controller. There seems to be
8 a need of various kinds of IO control policies (like proportional BW, max BW)
9 both at leaf nodes as well as at intermediate nodes in a storage hierarchy.
10 Plan is to use the same cgroup based management interface for blkio controller
11 and based on user options switch IO policies in the background.
13 One IO control policy is throttling policy which can be used to
14 specify upper IO rate limits on devices. This policy is implemented in
15 generic block layer and can be used on leaf nodes as well as higher
16 level logical devices like device mapper.
20 Throttling/Upper Limit policy
21 -----------------------------
22 - Enable Block IO controller::
26 - Enable throttling in block layer::
28 CONFIG_BLK_DEV_THROTTLING=y
30 - Mount blkio controller (see cgroups.txt, Why are cgroups needed?)::
32 mount -t cgroup -o blkio none /sys/fs/cgroup/blkio
34 - Specify a bandwidth rate on particular device for root group. The format
35 for policy is "<major>:<minor> <bytes_per_second>"::
37 echo "8:16 1048576" > /sys/fs/cgroup/blkio/blkio.throttle.read_bps_device
39 Above will put a limit of 1MB/second on reads happening for root group
40 on device having major/minor number 8:16.
42 - Run dd to read a file and see if rate is throttled to 1MB/s or not::
44 # dd iflag=direct if=/mnt/common/zerofile of=/dev/null bs=4K count=1024
47 4194304 bytes (4.2 MB) copied, 4.0001 s, 1.0 MB/s
49 Limits for writes can be put using blkio.throttle.write_bps_device file.
54 Throttling implements hierarchy support; however,
55 throttling's hierarchy support is enabled iff "sane_behavior" is
56 enabled from cgroup side, which currently is a development option and
57 not publicly available.
59 If somebody created a hierarchy like as follows::
67 Throttling with "sane_behavior" will handle the
68 hierarchy correctly. For throttling, all limits apply
69 to the whole subtree while all statistics are local to the IOs
70 directly generated by tasks in that cgroup.
72 Throttling without "sane_behavior" enabled from cgroup side will
73 practically treat all groups at same level as if it looks like the
78 root test1 test2 test3
80 Various user visible config options
81 ===================================
83 - Block IO controller.
85 CONFIG_BFQ_CGROUP_DEBUG
86 - Debug help. Right now some additional stats file show up in cgroup
87 if this option is enabled.
89 CONFIG_BLK_DEV_THROTTLING
90 - Enable block device throttling support in block layer.
92 Details of cgroup files
93 =======================
94 Proportional weight policy files
95 --------------------------------
97 - Specifies per cgroup weight. This is default weight of the group
98 on all the devices until and unless overridden by per device rule.
99 (See blkio.weight_device).
100 Currently allowed range of weights is from 10 to 1000.
102 - blkio.weight_device
103 - One can specify per cgroup per device rules using this interface.
104 These rules override the default value of group weight as specified
107 Following is the format::
109 # echo dev_maj:dev_minor weight > blkio.weight_device
111 Configure weight=300 on /dev/sdb (8:16) in this cgroup::
113 # echo 8:16 300 > blkio.weight_device
114 # cat blkio.weight_device
118 Configure weight=500 on /dev/sda (8:0) in this cgroup::
120 # echo 8:0 500 > blkio.weight_device
121 # cat blkio.weight_device
126 Remove specific weight for /dev/sda in this cgroup::
128 # echo 8:0 0 > blkio.weight_device
129 # cat blkio.weight_device
134 - disk time allocated to cgroup per device in milliseconds. First
135 two fields specify the major and minor number of the device and
136 third field specifies the disk time allocated to group in
140 - number of sectors transferred to/from disk by the group. First
141 two fields specify the major and minor number of the device and
142 third field specifies the number of sectors transferred by the
143 group to/from the device.
145 - blkio.io_service_bytes
146 - Number of bytes transferred to/from the disk by the group. These
147 are further divided by the type of operation - read or write, sync
148 or async. First two fields specify the major and minor number of the
149 device, third field specifies the operation type and the fourth field
150 specifies the number of bytes.
153 - Number of IOs (bio) issued to the disk by the group. These
154 are further divided by the type of operation - read or write, sync
155 or async. First two fields specify the major and minor number of the
156 device, third field specifies the operation type and the fourth field
157 specifies the number of IOs.
159 - blkio.io_service_time
160 - Total amount of time between request dispatch and request completion
161 for the IOs done by this cgroup. This is in nanoseconds to make it
162 meaningful for flash devices too. For devices with queue depth of 1,
163 this time represents the actual service time. When queue_depth > 1,
164 that is no longer true as requests may be served out of order. This
165 may cause the service time for a given IO to include the service time
166 of multiple IOs when served out of order which may result in total
167 io_service_time > actual time elapsed. This time is further divided by
168 the type of operation - read or write, sync or async. First two fields
169 specify the major and minor number of the device, third field
170 specifies the operation type and the fourth field specifies the
171 io_service_time in ns.
174 - Total amount of time the IOs for this cgroup spent waiting in the
175 scheduler queues for service. This can be greater than the total time
176 elapsed since it is cumulative io_wait_time for all IOs. It is not a
177 measure of total time the cgroup spent waiting but rather a measure of
178 the wait_time for its individual IOs. For devices with queue_depth > 1
179 this metric does not include the time spent waiting for service once
180 the IO is dispatched to the device but till it actually gets serviced
181 (there might be a time lag here due to re-ordering of requests by the
182 device). This is in nanoseconds to make it meaningful for flash
183 devices too. This time is further divided by the type of operation -
184 read or write, sync or async. First two fields specify the major and
185 minor number of the device, third field specifies the operation type
186 and the fourth field specifies the io_wait_time in ns.
189 - Total number of bios/requests merged into requests belonging to this
190 cgroup. This is further divided by the type of operation - read or
191 write, sync or async.
194 - Total number of requests queued up at any given instant for this
195 cgroup. This is further divided by the type of operation - read or
196 write, sync or async.
198 - blkio.avg_queue_size
199 - Debugging aid only enabled if CONFIG_BFQ_CGROUP_DEBUG=y.
200 The average queue size for this cgroup over the entire time of this
201 cgroup's existence. Queue size samples are taken each time one of the
202 queues of this cgroup gets a timeslice.
204 - blkio.group_wait_time
205 - Debugging aid only enabled if CONFIG_BFQ_CGROUP_DEBUG=y.
206 This is the amount of time the cgroup had to wait since it became busy
207 (i.e., went from 0 to 1 request queued) to get a timeslice for one of
208 its queues. This is different from the io_wait_time which is the
209 cumulative total of the amount of time spent by each IO in that cgroup
210 waiting in the scheduler queue. This is in nanoseconds. If this is
211 read when the cgroup is in a waiting (for timeslice) state, the stat
212 will only report the group_wait_time accumulated till the last time it
213 got a timeslice and will not include the current delta.
216 - Debugging aid only enabled if CONFIG_BFQ_CGROUP_DEBUG=y.
217 This is the amount of time a cgroup spends without any pending
218 requests when not being served, i.e., it does not include any time
219 spent idling for one of the queues of the cgroup. This is in
220 nanoseconds. If this is read when the cgroup is in an empty state,
221 the stat will only report the empty_time accumulated till the last
222 time it had a pending request and will not include the current delta.
225 - Debugging aid only enabled if CONFIG_BFQ_CGROUP_DEBUG=y.
226 This is the amount of time spent by the IO scheduler idling for a
227 given cgroup in anticipation of a better request than the existing ones
228 from other queues/cgroups. This is in nanoseconds. If this is read
229 when the cgroup is in an idling state, the stat will only report the
230 idle_time accumulated till the last idle period and will not include
234 - Debugging aid only enabled if CONFIG_BFQ_CGROUP_DEBUG=y. This
235 gives the statistics about how many a times a group was dequeued
236 from service tree of the device. First two fields specify the major
237 and minor number of the device and third field specifies the number
238 of times a group was dequeued from a particular device.
241 - Recursive version of various stats. These files show the
242 same information as their non-recursive counterparts but
243 include stats from all the descendant cgroups.
245 Throttling/Upper limit policy files
246 -----------------------------------
247 - blkio.throttle.read_bps_device
248 - Specifies upper limit on READ rate from the device. IO rate is
249 specified in bytes per second. Rules are per device. Following is
252 echo "<major>:<minor> <rate_bytes_per_second>" > /cgrp/blkio.throttle.read_bps_device
254 - blkio.throttle.write_bps_device
255 - Specifies upper limit on WRITE rate to the device. IO rate is
256 specified in bytes per second. Rules are per device. Following is
259 echo "<major>:<minor> <rate_bytes_per_second>" > /cgrp/blkio.throttle.write_bps_device
261 - blkio.throttle.read_iops_device
262 - Specifies upper limit on READ rate from the device. IO rate is
263 specified in IO per second. Rules are per device. Following is
266 echo "<major>:<minor> <rate_io_per_second>" > /cgrp/blkio.throttle.read_iops_device
268 - blkio.throttle.write_iops_device
269 - Specifies upper limit on WRITE rate to the device. IO rate is
270 specified in io per second. Rules are per device. Following is
273 echo "<major>:<minor> <rate_io_per_second>" > /cgrp/blkio.throttle.write_iops_device
275 Note: If both BW and IOPS rules are specified for a device, then IO is
276 subjected to both the constraints.
278 - blkio.throttle.io_serviced
279 - Number of IOs (bio) issued to the disk by the group. These
280 are further divided by the type of operation - read or write, sync
281 or async. First two fields specify the major and minor number of the
282 device, third field specifies the operation type and the fourth field
283 specifies the number of IOs.
285 - blkio.throttle.io_service_bytes
286 - Number of bytes transferred to/from the disk by the group. These
287 are further divided by the type of operation - read or write, sync
288 or async. First two fields specify the major and minor number of the
289 device, third field specifies the operation type and the fourth field
290 specifies the number of bytes.
292 Common files among various policies
293 -----------------------------------
295 - Writing an int to this file will result in resetting all the stats