| 1 | ===============================================
|
| 2 | Block layer statistics in /sys/block/<dev>/stat
|
| 3 | ===============================================
|
| 4 |
|
| 5 | This file documents the contents of the /sys/block/<dev>/stat file.
|
| 6 |
|
| 7 | The stat file provides several statistics about the state of block
|
| 8 | device <dev>.
|
| 9 |
|
| 10 | Q.
|
| 11 | Why are there multiple statistics in a single file? Doesn't sysfs
|
| 12 | normally contain a single value per file?
|
| 13 |
|
| 14 | A.
|
| 15 | By having a single file, the kernel can guarantee that the statistics
|
| 16 | represent a consistent snapshot of the state of the device. If the
|
| 17 | statistics were exported as multiple files containing one statistic
|
| 18 | each, it would be impossible to guarantee that a set of readings
|
| 19 | represent a single point in time.
|
| 20 |
|
| 21 | The stat file consists of a single line of text containing 17 decimal
|
| 22 | values separated by whitespace. The fields are summarized in the
|
| 23 | following table, and described in more detail below.
|
| 24 |
|
| 25 |
|
| 26 | =============== ============= =================================================
|
| 27 | Name units description
|
| 28 | =============== ============= =================================================
|
| 29 | read I/Os requests number of read I/Os processed
|
| 30 | read merges requests number of read I/Os merged with in-queue I/O
|
| 31 | read sectors sectors number of sectors read
|
| 32 | read ticks milliseconds total wait time for read requests
|
| 33 | write I/Os requests number of write I/Os processed
|
| 34 | write merges requests number of write I/Os merged with in-queue I/O
|
| 35 | write sectors sectors number of sectors written
|
| 36 | write ticks milliseconds total wait time for write requests
|
| 37 | in_flight requests number of I/Os currently in flight
|
| 38 | io_ticks milliseconds total time this block device has been active
|
| 39 | time_in_queue milliseconds total wait time for all requests
|
| 40 | discard I/Os requests number of discard I/Os processed
|
| 41 | discard merges requests number of discard I/Os merged with in-queue I/O
|
| 42 | discard sectors sectors number of sectors discarded
|
| 43 | discard ticks milliseconds total wait time for discard requests
|
| 44 | flush I/Os requests number of flush I/Os processed
|
| 45 | flush ticks milliseconds total wait time for flush requests
|
| 46 | =============== ============= =================================================
|
| 47 |
|
| 48 | read I/Os, write I/Os, discard I/0s
|
| 49 | ===================================
|
| 50 |
|
| 51 | These values increment when an I/O request completes.
|
| 52 |
|
| 53 | flush I/Os
|
| 54 | ==========
|
| 55 |
|
| 56 | These values increment when an flush I/O request completes.
|
| 57 |
|
| 58 | Block layer combines flush requests and executes at most one at a time.
|
| 59 | This counts flush requests executed by disk. Not tracked for partitions.
|
| 60 |
|
| 61 | read merges, write merges, discard merges
|
| 62 | =========================================
|
| 63 |
|
| 64 | These values increment when an I/O request is merged with an
|
| 65 | already-queued I/O request.
|
| 66 |
|
| 67 | read sectors, write sectors, discard_sectors
|
| 68 | ============================================
|
| 69 |
|
| 70 | These values count the number of sectors read from, written to, or
|
| 71 | discarded from this block device. The "sectors" in question are the
|
| 72 | standard UNIX 512-byte sectors, not any device- or filesystem-specific
|
| 73 | block size. The counters are incremented when the I/O completes.
|
| 74 |
|
| 75 | read ticks, write ticks, discard ticks, flush ticks
|
| 76 | ===================================================
|
| 77 |
|
| 78 | These values count the number of milliseconds that I/O requests have
|
| 79 | waited on this block device. If there are multiple I/O requests waiting,
|
| 80 | these values will increase at a rate greater than 1000/second; for
|
| 81 | example, if 60 read requests wait for an average of 30 ms, the read_ticks
|
| 82 | field will increase by 60*30 = 1800.
|
| 83 |
|
| 84 | in_flight
|
| 85 | =========
|
| 86 |
|
| 87 | This value counts the number of I/O requests that have been issued to
|
| 88 | the device driver but have not yet completed. It does not include I/O
|
| 89 | requests that are in the queue but not yet issued to the device driver.
|
| 90 |
|
| 91 | io_ticks
|
| 92 | ========
|
| 93 |
|
| 94 | This value counts the number of milliseconds during which the device has
|
| 95 | had I/O requests queued.
|
| 96 |
|
| 97 | time_in_queue
|
| 98 | =============
|
| 99 |
|
| 100 | This value counts the number of milliseconds that I/O requests have waited
|
| 101 | on this block device. If there are multiple I/O requests waiting, this
|
| 102 | value will increase as the product of the number of milliseconds times the
|
| 103 | number of requests waiting (see "read ticks" above for an example).
|