blob: c76c21d87e8582a2bd624ac126ba45290e851d70 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001I/O statistics fields
2---------------
3
Linus Torvalds1da177e2005-04-16 15:20:36 -07004Since 2.4.20 (and some versions before, with patches), and 2.5.45,
5more extensive disk statistics have been introduced to help measure disk
6activity. Tools such as sar and iostat typically interpret these and do
7the work for you, but in case you are interested in creating your own
8tools, the fields are explained here.
9
10In 2.4 now, the information is found as additional fields in
11/proc/partitions. In 2.6, the same information is found in two
12places: one is in the file /proc/diskstats, and the other is within
13the sysfs file system, which must be mounted in order to obtain
14the information. Throughout this document we'll assume that sysfs
15is mounted on /sys, although of course it may be mounted anywhere.
16Both /proc/diskstats and sysfs use the same source for the information
17and so should not differ.
18
19Here are examples of these different formats:
20
212.4:
22 3 0 39082680 hda 446216 784926 9550688 4382310 424847 312726 5922052 19310380 0 3376340 23705160
23 3 1 9221278 hda1 35486 0 35496 38030 0 0 0 0 0 38030 38030
24
25
262.6 sysfs:
27 446216 784926 9550688 4382310 424847 312726 5922052 19310380 0 3376340 23705160
28 35486 38030 38030 38030
29
302.6 diskstats:
31 3 0 hda 446216 784926 9550688 4382310 424847 312726 5922052 19310380 0 3376340 23705160
32 3 1 hda1 35486 38030 38030 38030
33
34On 2.4 you might execute "grep 'hda ' /proc/partitions". On 2.6, you have
35a choice of "cat /sys/block/hda/stat" or "grep 'hda ' /proc/diskstats".
36The advantage of one over the other is that the sysfs choice works well
37if you are watching a known, small set of disks. /proc/diskstats may
38be a better choice if you are watching a large number of disks because
39you'll avoid the overhead of 50, 100, or 500 or more opens/closes with
40each snapshot of your disk statistics.
41
42In 2.4, the statistics fields are those after the device name. In
43the above example, the first field of statistics would be 446216.
44By contrast, in 2.6 if you look at /sys/block/hda/stat, you'll
45find just the eleven fields, beginning with 446216. If you look at
46/proc/diskstats, the eleven fields will be preceded by the major and
Randy Dunlap9d2e1572011-03-23 20:44:18 +010047minor device numbers, and device name. Each of these formats provides
Linus Torvalds1da177e2005-04-16 15:20:36 -070048eleven fields of statistics, each meaning exactly the same things.
49All fields except field 9 are cumulative since boot. Field 9 should
Randy Dunlap9d2e1572011-03-23 20:44:18 +010050go to zero as I/Os complete; all others only increase (unless they
51overflow and wrap). Yes, these are (32-bit or 64-bit) unsigned long
52(native word size) numbers, and on a very busy or long-lived system they
Linus Torvalds1da177e2005-04-16 15:20:36 -070053may wrap. Applications should be prepared to deal with that; unless
54your observations are measured in large numbers of minutes or hours,
55they should not wrap twice before you notice them.
56
57Each set of stats only applies to the indicated device; if you want
58system-wide stats you'll have to find all the devices and sum them all up.
59
Jerome Marchand0e53c2b2008-02-08 11:10:56 +010060Field 1 -- # of reads completed
Linus Torvalds1da177e2005-04-16 15:20:36 -070061 This is the total number of reads completed successfully.
62Field 2 -- # of reads merged, field 6 -- # of writes merged
63 Reads and writes which are adjacent to each other may be merged for
64 efficiency. Thus two 4K reads may become one 8K read before it is
65 ultimately handed to the disk, and so it will be counted (and queued)
66 as only one I/O. This field lets you know how often this was done.
67Field 3 -- # of sectors read
68 This is the total number of sectors read successfully.
69Field 4 -- # of milliseconds spent reading
70 This is the total number of milliseconds spent by all reads (as
71 measured from __make_request() to end_that_request_last()).
72Field 5 -- # of writes completed
73 This is the total number of writes completed successfully.
74Field 7 -- # of sectors written
75 This is the total number of sectors written successfully.
76Field 8 -- # of milliseconds spent writing
77 This is the total number of milliseconds spent by all writes (as
78 measured from __make_request() to end_that_request_last()).
79Field 9 -- # of I/Os currently in progress
80 The only field that should go to zero. Incremented as requests are
Jens Axboe165125e2007-07-24 09:28:11 +020081 given to appropriate struct request_queue and decremented as they finish.
Linus Torvalds1da177e2005-04-16 15:20:36 -070082Field 10 -- # of milliseconds spent doing I/Os
Jim Cromie50ed3802010-07-03 23:18:11 -060083 This field increases so long as field 9 is nonzero.
Linus Torvalds1da177e2005-04-16 15:20:36 -070084Field 11 -- weighted # of milliseconds spent doing I/Os
85 This field is incremented at each I/O start, I/O completion, I/O
86 merge, or read of these stats by the number of I/Os in progress
87 (field 9) times the number of milliseconds spent doing I/O since the
88 last update of this field. This can provide an easy measure of both
89 I/O completion time and the backlog that may be accumulating.
90
91
92To avoid introducing performance bottlenecks, no locks are held while
93modifying these counters. This implies that minor inaccuracies may be
94introduced when changes collide, so (for instance) adding up all the
95read I/Os issued per partition should equal those made to the disks ...
96but due to the lack of locking it may only be very close.
97
Randy Dunlap9d2e1572011-03-23 20:44:18 +010098In 2.6, there are counters for each CPU, which make the lack of locking
99almost a non-issue. When the statistics are read, the per-CPU counters
100are summed (possibly overflowing the unsigned long variable they are
Linus Torvalds1da177e2005-04-16 15:20:36 -0700101summed to) and the result given to the user. There is no convenient
Randy Dunlap9d2e1572011-03-23 20:44:18 +0100102user interface for accessing the per-CPU counters themselves.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103
104Disks vs Partitions
105-------------------
106
107There were significant changes between 2.4 and 2.6 in the I/O subsystem.
108As a result, some statistic information disappeared. The translation from
109a disk address relative to a partition to the disk address relative to
110the host disk happens much earlier. All merges and timings now happen
111at the disk level rather than at both the disk and partition level as
112in 2.4. Consequently, you'll see a different statistics output on 2.6 for
113partitions from that for disks. There are only *four* fields available
114for partitions on 2.6 machines. This is reflected in the examples above.
115
116Field 1 -- # of reads issued
117 This is the total number of reads issued to this partition.
118Field 2 -- # of sectors read
119 This is the total number of sectors requested to be read from this
120 partition.
121Field 3 -- # of writes issued
122 This is the total number of writes issued to this partition.
123Field 4 -- # of sectors written
124 This is the total number of sectors requested to be written to
125 this partition.
126
127Note that since the address is translated to a disk-relative one, and no
128record of the partition-relative address is kept, the subsequent success
129or failure of the read cannot be attributed to the partition. In other
130words, the number of reads for partitions is counted slightly before time
131of queuing for partitions, and at completion for whole disks. This is
132a subtle distinction that is probably uninteresting for most cases.
133
Jerome Marchand0e53c2b2008-02-08 11:10:56 +0100134More significant is the error induced by counting the numbers of
135reads/writes before merges for partitions and after for disks. Since a
136typical workload usually contains a lot of successive and adjacent requests,
137the number of reads/writes issued can be several times higher than the
138number of reads/writes completed.
139
140In 2.6.25, the full statistic set is again available for partitions and
141disk and partition statistics are consistent again. Since we still don't
142keep record of the partition-relative address, an operation is attributed to
143the partition which contains the first sector of the request after the
144eventual merges. As requests can be merged across partition, this could lead
Matt LaPlanted9195882008-07-25 19:45:33 -0700145to some (probably insignificant) inaccuracy.
Jerome Marchand0e53c2b2008-02-08 11:10:56 +0100146
Linus Torvalds1da177e2005-04-16 15:20:36 -0700147Additional notes
148----------------
149
150In 2.6, sysfs is not mounted by default. If your distribution of
151Linux hasn't added it already, here's the line you'll want to add to
152your /etc/fstab:
153
154none /sys sysfs defaults 0 0
155
156
157In 2.6, all disk statistics were removed from /proc/stat. In 2.4, they
158appear in both /proc/partitions and /proc/stat, although the ones in
159/proc/stat take a very different format from those in /proc/partitions
160(see proc(5), if your system has it.)
161
162-- ricklind@us.ibm.com