aboutsummaryrefslogtreecommitdiff
path: root/Documentation/iostats.txt
diff options
context:
space:
mode:
authorFathi Boudra <fathi.boudra@linaro.org>2013-04-28 09:33:08 +0300
committerFathi Boudra <fathi.boudra@linaro.org>2013-04-28 09:33:08 +0300
commit3b4bd47f8f4ed3aaf7c81c9b5d2d37ad79fadf4a (patch)
treeb9996006addfd7ae70a39672b76843b49aebc189 /Documentation/iostats.txt
downloadlinux-linaro-highbank-3b4bd47f8f4ed3aaf7c81c9b5d2d37ad79fadf4a.tar.gz
Imported Upstream version 3.9.0HEADupstream/3.9.0upstreammaster
Diffstat (limited to 'Documentation/iostats.txt')
-rw-r--r--Documentation/iostats.txt162
1 files changed, 162 insertions, 0 deletions
diff --git a/Documentation/iostats.txt b/Documentation/iostats.txt
new file mode 100644
index 00000000..c76c21d8
--- /dev/null
+++ b/Documentation/iostats.txt
@@ -0,0 +1,162 @@
+I/O statistics fields
+---------------
+
+Since 2.4.20 (and some versions before, with patches), and 2.5.45,
+more extensive disk statistics have been introduced to help measure disk
+activity. Tools such as sar and iostat typically interpret these and do
+the work for you, but in case you are interested in creating your own
+tools, the fields are explained here.
+
+In 2.4 now, the information is found as additional fields in
+/proc/partitions. In 2.6, the same information is found in two
+places: one is in the file /proc/diskstats, and the other is within
+the sysfs file system, which must be mounted in order to obtain
+the information. Throughout this document we'll assume that sysfs
+is mounted on /sys, although of course it may be mounted anywhere.
+Both /proc/diskstats and sysfs use the same source for the information
+and so should not differ.
+
+Here are examples of these different formats:
+
+2.4:
+ 3 0 39082680 hda 446216 784926 9550688 4382310 424847 312726 5922052 19310380 0 3376340 23705160
+ 3 1 9221278 hda1 35486 0 35496 38030 0 0 0 0 0 38030 38030
+
+
+2.6 sysfs:
+ 446216 784926 9550688 4382310 424847 312726 5922052 19310380 0 3376340 23705160
+ 35486 38030 38030 38030
+
+2.6 diskstats:
+ 3 0 hda 446216 784926 9550688 4382310 424847 312726 5922052 19310380 0 3376340 23705160
+ 3 1 hda1 35486 38030 38030 38030
+
+On 2.4 you might execute "grep 'hda ' /proc/partitions". On 2.6, you have
+a choice of "cat /sys/block/hda/stat" or "grep 'hda ' /proc/diskstats".
+The advantage of one over the other is that the sysfs choice works well
+if you are watching a known, small set of disks. /proc/diskstats may
+be a better choice if you are watching a large number of disks because
+you'll avoid the overhead of 50, 100, or 500 or more opens/closes with
+each snapshot of your disk statistics.
+
+In 2.4, the statistics fields are those after the device name. In
+the above example, the first field of statistics would be 446216.
+By contrast, in 2.6 if you look at /sys/block/hda/stat, you'll
+find just the eleven fields, beginning with 446216. If you look at
+/proc/diskstats, the eleven fields will be preceded by the major and
+minor device numbers, and device name. Each of these formats provides
+eleven fields of statistics, each meaning exactly the same things.
+All fields except field 9 are cumulative since boot. Field 9 should
+go to zero as I/Os complete; all others only increase (unless they
+overflow and wrap). Yes, these are (32-bit or 64-bit) unsigned long
+(native word size) numbers, and on a very busy or long-lived system they
+may wrap. Applications should be prepared to deal with that; unless
+your observations are measured in large numbers of minutes or hours,
+they should not wrap twice before you notice them.
+
+Each set of stats only applies to the indicated device; if you want
+system-wide stats you'll have to find all the devices and sum them all up.
+
+Field 1 -- # of reads completed
+ This is the total number of reads completed successfully.
+Field 2 -- # of reads merged, field 6 -- # of writes merged
+ Reads and writes which are adjacent to each other may be merged for
+ efficiency. Thus two 4K reads may become one 8K read before it is
+ ultimately handed to the disk, and so it will be counted (and queued)
+ as only one I/O. This field lets you know how often this was done.
+Field 3 -- # of sectors read
+ This is the total number of sectors read successfully.
+Field 4 -- # of milliseconds spent reading
+ This is the total number of milliseconds spent by all reads (as
+ measured from __make_request() to end_that_request_last()).
+Field 5 -- # of writes completed
+ This is the total number of writes completed successfully.
+Field 7 -- # of sectors written
+ This is the total number of sectors written successfully.
+Field 8 -- # of milliseconds spent writing
+ This is the total number of milliseconds spent by all writes (as
+ measured from __make_request() to end_that_request_last()).
+Field 9 -- # of I/Os currently in progress
+ The only field that should go to zero. Incremented as requests are
+ given to appropriate struct request_queue and decremented as they finish.
+Field 10 -- # of milliseconds spent doing I/Os
+ This field increases so long as field 9 is nonzero.
+Field 11 -- weighted # of milliseconds spent doing I/Os
+ This field is incremented at each I/O start, I/O completion, I/O
+ merge, or read of these stats by the number of I/Os in progress
+ (field 9) times the number of milliseconds spent doing I/O since the
+ last update of this field. This can provide an easy measure of both
+ I/O completion time and the backlog that may be accumulating.
+
+
+To avoid introducing performance bottlenecks, no locks are held while
+modifying these counters. This implies that minor inaccuracies may be
+introduced when changes collide, so (for instance) adding up all the
+read I/Os issued per partition should equal those made to the disks ...
+but due to the lack of locking it may only be very close.
+
+In 2.6, there are counters for each CPU, which make the lack of locking
+almost a non-issue. When the statistics are read, the per-CPU counters
+are summed (possibly overflowing the unsigned long variable they are
+summed to) and the result given to the user. There is no convenient
+user interface for accessing the per-CPU counters themselves.
+
+Disks vs Partitions
+-------------------
+
+There were significant changes between 2.4 and 2.6 in the I/O subsystem.
+As a result, some statistic information disappeared. The translation from
+a disk address relative to a partition to the disk address relative to
+the host disk happens much earlier. All merges and timings now happen
+at the disk level rather than at both the disk and partition level as
+in 2.4. Consequently, you'll see a different statistics output on 2.6 for
+partitions from that for disks. There are only *four* fields available
+for partitions on 2.6 machines. This is reflected in the examples above.
+
+Field 1 -- # of reads issued
+ This is the total number of reads issued to this partition.
+Field 2 -- # of sectors read
+ This is the total number of sectors requested to be read from this
+ partition.
+Field 3 -- # of writes issued
+ This is the total number of writes issued to this partition.
+Field 4 -- # of sectors written
+ This is the total number of sectors requested to be written to
+ this partition.
+
+Note that since the address is translated to a disk-relative one, and no
+record of the partition-relative address is kept, the subsequent success
+or failure of the read cannot be attributed to the partition. In other
+words, the number of reads for partitions is counted slightly before time
+of queuing for partitions, and at completion for whole disks. This is
+a subtle distinction that is probably uninteresting for most cases.
+
+More significant is the error induced by counting the numbers of
+reads/writes before merges for partitions and after for disks. Since a
+typical workload usually contains a lot of successive and adjacent requests,
+the number of reads/writes issued can be several times higher than the
+number of reads/writes completed.
+
+In 2.6.25, the full statistic set is again available for partitions and
+disk and partition statistics are consistent again. Since we still don't
+keep record of the partition-relative address, an operation is attributed to
+the partition which contains the first sector of the request after the
+eventual merges. As requests can be merged across partition, this could lead
+to some (probably insignificant) inaccuracy.
+
+Additional notes
+----------------
+
+In 2.6, sysfs is not mounted by default. If your distribution of
+Linux hasn't added it already, here's the line you'll want to add to
+your /etc/fstab:
+
+none /sys sysfs defaults 0 0
+
+
+In 2.6, all disk statistics were removed from /proc/stat. In 2.4, they
+appear in both /proc/partitions and /proc/stat, although the ones in
+/proc/stat take a very different format from those in /proc/partitions
+(see proc(5), if your system has it.)
+
+-- ricklind@us.ibm.com