211 lines · plain
1=====================2I/O statistics fields3=====================4 5Since 2.4.20 (and some versions before, with patches), and 2.5.45,6more extensive disk statistics have been introduced to help measure disk7activity. Tools such as ``sar`` and ``iostat`` typically interpret these and do8the work for you, but in case you are interested in creating your own9tools, the fields are explained here.10 11In 2.4 now, the information is found as additional fields in12``/proc/partitions``. In 2.6 and upper, the same information is found in two13places: one is in the file ``/proc/diskstats``, and the other is within14the sysfs file system, which must be mounted in order to obtain15the information. Throughout this document we'll assume that sysfs16is mounted on ``/sys``, although of course it may be mounted anywhere.17Both ``/proc/diskstats`` and sysfs use the same source for the information18and so should not differ.19 20Here are examples of these different formats::21 22 2.4:23 3 0 39082680 hda 446216 784926 9550688 4382310 424847 312726 5922052 19310380 0 3376340 2370516024 3 1 9221278 hda1 35486 0 35496 38030 0 0 0 0 0 38030 3803025 26 2.6+ sysfs:27 446216 784926 9550688 4382310 424847 312726 5922052 19310380 0 3376340 2370516028 35486 38030 38030 3803029 30 2.6+ diskstats:31 3 0 hda 446216 784926 9550688 4382310 424847 312726 5922052 19310380 0 3376340 2370516032 3 1 hda1 35486 38030 38030 3803033 34 4.18+ diskstats:35 3 0 hda 446216 784926 9550688 4382310 424847 312726 5922052 19310380 0 3376340 23705160 0 0 0 036 37On 2.4 you might execute ``grep 'hda ' /proc/partitions``. On 2.6+, you have38a choice of ``cat /sys/block/hda/stat`` or ``grep 'hda ' /proc/diskstats``.39 40The advantage of one over the other is that the sysfs choice works well41if you are watching a known, small set of disks. ``/proc/diskstats`` may42be a better choice if you are watching a large number of disks because43you'll avoid the overhead of 50, 100, or 500 or more opens/closes with44each snapshot of your disk statistics.45 46In 2.4, the statistics fields are those after the device name. In47the above example, the first field of statistics would be 446216.48By contrast, in 2.6+ if you look at ``/sys/block/hda/stat``, you'll49find just the 15 fields, beginning with 446216. If you look at50``/proc/diskstats``, the 15 fields will be preceded by the major and51minor device numbers, and device name. Each of these formats provides5215 fields of statistics, each meaning exactly the same things.53All fields except field 9 are cumulative since boot. Field 9 should54go to zero as I/Os complete; all others only increase (unless they55overflow and wrap). Wrapping might eventually occur on a very busy56or long-lived system; so applications should be prepared to deal with57it. Regarding wrapping, the types of the fields are either unsigned58int (32 bit) or unsigned long (32-bit or 64-bit, depending on your59machine) as noted per-field below. Unless your observations are very60spread in time, these fields should not wrap twice before you notice it.61 62Each set of stats only applies to the indicated device; if you want63system-wide stats you'll have to find all the devices and sum them all up.64 65Field 1 -- # of reads completed (unsigned long)66 This is the total number of reads completed successfully.67 68Field 2 -- # of reads merged, field 6 -- # of writes merged (unsigned long)69 Reads and writes which are adjacent to each other may be merged for70 efficiency. Thus two 4K reads may become one 8K read before it is71 ultimately handed to the disk, and so it will be counted (and queued)72 as only one I/O. This field lets you know how often this was done.73 74Field 3 -- # of sectors read (unsigned long)75 This is the total number of sectors read successfully.76 77Field 4 -- # of milliseconds spent reading (unsigned int)78 This is the total number of milliseconds spent by all reads (as79 measured from blk_mq_alloc_request() to __blk_mq_end_request()).80 81Field 5 -- # of writes completed (unsigned long)82 This is the total number of writes completed successfully.83 84Field 6 -- # of writes merged (unsigned long)85 See the description of field 2.86 87Field 7 -- # of sectors written (unsigned long)88 This is the total number of sectors written successfully.89 90Field 8 -- # of milliseconds spent writing (unsigned int)91 This is the total number of milliseconds spent by all writes (as92 measured from blk_mq_alloc_request() to __blk_mq_end_request()).93 94Field 9 -- # of I/Os currently in progress (unsigned int)95 The only field that should go to zero. Incremented as requests are96 given to appropriate struct request_queue and decremented as they finish.97 98Field 10 -- # of milliseconds spent doing I/Os (unsigned int)99 This field increases so long as field 9 is nonzero.100 101 Since 5.0 this field counts jiffies when at least one request was102 started or completed. If request runs more than 2 jiffies then some103 I/O time might be not accounted in case of concurrent requests.104 105Field 11 -- weighted # of milliseconds spent doing I/Os (unsigned int)106 This field is incremented at each I/O start, I/O completion, I/O107 merge, or read of these stats by the number of I/Os in progress108 (field 9) times the number of milliseconds spent doing I/O since the109 last update of this field. This can provide an easy measure of both110 I/O completion time and the backlog that may be accumulating.111 112Field 12 -- # of discards completed (unsigned long)113 This is the total number of discards completed successfully.114 115Field 13 -- # of discards merged (unsigned long)116 See the description of field 2117 118Field 14 -- # of sectors discarded (unsigned long)119 This is the total number of sectors discarded successfully.120 121Field 15 -- # of milliseconds spent discarding (unsigned int)122 This is the total number of milliseconds spent by all discards (as123 measured from blk_mq_alloc_request() to __blk_mq_end_request()).124 125Field 16 -- # of flush requests completed126 This is the total number of flush requests completed successfully.127 128 Block layer combines flush requests and executes at most one at a time.129 This counts flush requests executed by disk. Not tracked for partitions.130 131Field 17 -- # of milliseconds spent flushing132 This is the total number of milliseconds spent by all flush requests.133 134To avoid introducing performance bottlenecks, no locks are held while135modifying these counters. This implies that minor inaccuracies may be136introduced when changes collide, so (for instance) adding up all the137read I/Os issued per partition should equal those made to the disks ...138but due to the lack of locking it may only be very close.139 140In 2.6+, there are counters for each CPU, which make the lack of locking141almost a non-issue. When the statistics are read, the per-CPU counters142are summed (possibly overflowing the unsigned long variable they are143summed to) and the result given to the user. There is no convenient144user interface for accessing the per-CPU counters themselves.145 146Since 4.19 request times are measured with nanoseconds precision and147truncated to milliseconds before showing in this interface.148 149Disks vs Partitions150-------------------151 152There were significant changes between 2.4 and 2.6+ in the I/O subsystem.153As a result, some statistic information disappeared. The translation from154a disk address relative to a partition to the disk address relative to155the host disk happens much earlier. All merges and timings now happen156at the disk level rather than at both the disk and partition level as157in 2.4. Consequently, you'll see a different statistics output on 2.6+ for158partitions from that for disks. There are only *four* fields available159for partitions on 2.6+ machines. This is reflected in the examples above.160 161Field 1 -- # of reads issued162 This is the total number of reads issued to this partition.163 164Field 2 -- # of sectors read165 This is the total number of sectors requested to be read from this166 partition.167 168Field 3 -- # of writes issued169 This is the total number of writes issued to this partition.170 171Field 4 -- # of sectors written172 This is the total number of sectors requested to be written to173 this partition.174 175Note that since the address is translated to a disk-relative one, and no176record of the partition-relative address is kept, the subsequent success177or failure of the read cannot be attributed to the partition. In other178words, the number of reads for partitions is counted slightly before time179of queuing for partitions, and at completion for whole disks. This is180a subtle distinction that is probably uninteresting for most cases.181 182More significant is the error induced by counting the numbers of183reads/writes before merges for partitions and after for disks. Since a184typical workload usually contains a lot of successive and adjacent requests,185the number of reads/writes issued can be several times higher than the186number of reads/writes completed.187 188In 2.6.25, the full statistic set is again available for partitions and189disk and partition statistics are consistent again. Since we still don't190keep record of the partition-relative address, an operation is attributed to191the partition which contains the first sector of the request after the192eventual merges. As requests can be merged across partition, this could lead193to some (probably insignificant) inaccuracy.194 195Additional notes196----------------197 198In 2.6+, sysfs is not mounted by default. If your distribution of199Linux hasn't added it already, here's the line you'll want to add to200your ``/etc/fstab``::201 202 none /sys sysfs defaults 0 0203 204 205In 2.6+, all disk statistics were removed from ``/proc/stat``. In 2.4, they206appear in both ``/proc/partitions`` and ``/proc/stat``, although the ones in207``/proc/stat`` take a very different format from those in ``/proc/partitions``208(see proc(5), if your system has it.)209 210-- ricklind@us.ibm.com211