Skip to content

Disk IO

Diskio collector is used to collect the index of disk flow and time.

Configuration

After successfully installing and starting DataKit, the DiskIO collector will be enabled by default without the need for manual activation.

Precondition

For some older versions of Windows operating systems, if you encounter an error with Datakit: "The system cannot find the file specified."

Run PowerShell as an administrator and execute:

diskperf -Y

The Datakit service needs to be restarted after successful execution.

Collector Configuration

Go to the conf.d/host directory under the DataKit installation directory, copy diskio.conf.sample and name it diskio.conf. Examples are as follows:

[[inputs.diskio]]
  ##(optional) collect interval, default is 10 seconds
  interval = '10s'

  ## By default, gather stats for all devices including
  ## disk partitions.
  ## Setting interfaces using regular expressions will collect these expected devices.
  # devices = ['''^sda\d*''', '''^sdb\d*''', '''vd.*''']

  ## If the disk serial number is not required, please uncomment the following line.
  # skip_serial_number = true

  ## On systems which support it, device metadata can be added in the form of
  ## tags.
  ## Currently only Linux is supported via udev properties. You can view
  ## available properties for a device by running:
  ## 'udevadm info -q property -n /dev/sda'
  ## Note: Most, but not all, udev properties can be accessed this way. Properties
  ## that are currently inaccessible include DEVTYPE, DEVNAME, and DEVPATH.
  # device_tags = ["ID_FS_TYPE", "ID_FS_USAGE"]

  ## Using the same metadata source as device_tags,
  ## you can also customize the name of the device through a template.
  ## The "name_templates" parameter is a list of templates to try to apply equipment.
  ## The template can contain variables of the form "$PROPERTY" or "${PROPERTY}".
  ## The first template that does not contain any variables that do not exist
  ## for the device is used as the device name label.
  ## A typical use case for LVM volumes is to obtain VG/LV names,
  ## not DM-0 names which are almost meaningless.
  ## In addition, "device" is reserved specifically to indicate the device name.
  # name_templates = ["$ID_FS_LABEL","$DM_VG_NAME/$DM_LV_NAME", "$device:$ID_FS_TYPE"]

[inputs.diskio.tags]
  # some_tag = "some_value"
  # more_tag = "some_other_value"

Once configured, restart DataKit.

Can be turned on by ConfigMap Injection Collector Configuration or Config ENV_DATAKIT_INPUTS .

Can also be turned on by environment variables, (needs to be added as the default collector in ENV_DEFAULT_ENABLED_INPUTS):

  • ENV_INPUT_DISKIO_INTERVAL

    Collect interval

    Type: Duration

    input.conf: interval

    Default: 10s

  • ENV_INPUT_DISKIO_DEVICES

    Setting interfaces using regular expressions will collect these expected devices

    Type: List

    input.conf: devices

    Example: ^sda\d,^sdb\d,vd.*

  • ENV_INPUT_DISKIO_DEVICE_TAGS

    Device metadata added tags

    Type: List

    input.conf: device_tags

    Example: ID_FS_TYPE,ID_FS_USAGE

  • ENV_INPUT_DISKIO_NAME_TEMPLATES

    Using the same metadata source as device_tags

    Type: List

    input.conf: name_templates

    Example: $ID_FS_LABEL,$DM_VG_NAME/$DM_LV_NAME

  • ENV_INPUT_DISKIO_SKIP_SERIAL_NUMBER

    disk serial number is not required

    Type: Boolean

    input.conf: skip_serial_number

    Default: false

  • ENV_INPUT_DISKIO_TAGS

    Customize tags. If there is a tag with the same name in the configuration file, it will be overwritten

    Type: Map

    input.conf: tags

    Example: tag1=value1,tag2=value2

Metric

For all of the following data collections, a global tag named host is appended by default (the tag value is the host name of the DataKit), or it can be named by [[inputs.diskio.tags]] alternative host in the configuration.

diskio

  • tag
Tag Description
host System hostname.
name Device name.
  • metric list
Metric Description Type Unit
io_time Time spent doing I/Os. int ms
iops_in_progress I/Os currently in progress. int count
merged_reads The number of merged read requests. int count
merged_writes The number of merged write requests. int count
read_bytes The number of bytes read from the device. int B
read_bytes/sec The number of bytes read from the per second. int B/S
read_time Time spent reading. int ms
reads The number of read requests. int count
weighted_io_time Weighted time spent doing I/Os. int ms
write_bytes The number of bytes written to the device. int B
write_bytes/sec The number of bytes written to the device per second. int B/S
write_time Time spent writing. int ms
writes The number of write requests. int count

Extended Metric

Version-1.5.7

Collecting disk await for Linux

By default, DataKit cannot collect the disk await metric. If you need to obtain this metric, you can collect it by Custom Collector with Python.

Preconditions

Enter the DataKit installation directory, copy the pythond.conf.sample file and rename it to pythond.conf. Modify the corresponding configuration as follows:

[[inputs.pythond]]

    # Python collector name 
    name = 'diskio'  # required

    # Environment variables 
    #envs = ['LD_LIBRARY_PATH=/path/to/lib:$LD_LIBRARY_PATH',]

    # Python collector executable path (preferably use absolute path) 
    cmd = "python3" # required. python3 is recommended.

    # Relative path of the user script
    dirs = ["diskio"]

Install from Ubuntu

sudo apt-get install sysstat

sudo vi /etc/default/sysstat
# change ENABLED="false" to ENABLED="true"

sudo service sysstat restart

After installation, you can ran the following command to check if it was successful.

sar -d -p 3 1

Linux 2.6.32-696.el6.x86_64 (lgh)   10/06/2019      _x86_64_        (32 CPU)

10:08:16 PM       DEV       tps  rd_sec/s  wr_sec/s  avgrq-sz  avgqu-sz     await     svctm     %util
10:08:17 PM    dev8-0      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
10:08:17 PM  dev253-0      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
10:08:17 PM  dev253-1      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00

10:08:17 PM       DEV       tps  rd_sec/s  wr_sec/s  avgrq-sz  avgqu-sz     await     svctm     %util
10:08:18 PM    dev8-0      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
10:08:18 PM  dev253-0      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
10:08:18 PM  dev253-1      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00

10:08:18 PM       DEV       tps  rd_sec/s  wr_sec/s  avgrq-sz  avgqu-sz     await     svctm     %util
10:08:19 PM    dev8-0      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
10:08:19 PM  dev253-0      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
10:08:19 PM  dev253-1      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00

Average:          DEV       tps  rd_sec/s  wr_sec/s  avgrq-sz  avgqu-sz     await     svctm     %util
Average:       dev8-0      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
Average:     dev253-0      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
Average:     dev253-1      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00

Collect Scripts

Create file <DataKit Dir>/python.d/diskio/diskio.py and add the following content:

import subprocess
import re
from datakit_framework import DataKitFramework


class DiskIO(DataKitFramework):
    name = "diskio"
    interval = 10

    def run(self):
        stats = self.getStats()

        data = []

        for s in stats:
            tags = {
                "name": s.get("DEV", "")
            }
            awaitVal = 0.0
            svctmVal = 0.0

            try:
                awaitVal = float(s.get("await"))
            except:
                awaitVal = 0.0
            try:
                svctmVal = float(s.get("svctm"))
            except:
                svctmVal = 0.0

            fields = {
                "await": awaitVal,
                "svctm": svctmVal
            }
            data.append({
                "measurement": "diskio",
                "tags": tags,
                "fields": fields
            })

        in_data = {
            "M": data,
            "input": "datakitpy"
        }

        return self.report(in_data)

    def getStats(self):
        result = subprocess.run(
            ["sar", "-d", "-p", "3", "1"], stdout=subprocess.PIPE)
        output = result.stdout.decode("utf-8")

        str_list = output.splitlines()

        columns = []
        stats = []
        pattern = r'\s+'
        isAverage = False
        for l in enumerate(str_list):
            index, content = l
            if index < 2:
                continue

            stat = re.split(pattern, content)

            if len(stat) == 0 or stat[0] == "":
                isAverage = True
                continue

            if not isAverage:
                continue
            if "await" in stat and "DEV" in stat:
                columns = stat
            else:
                stat_info = {}
                if len(stat) != len(columns):
                    continue

                for s in enumerate(columns):
                    index, name = s
                    if index == 0:
                        continue
                    stat_info[name] = stat[index]
                stats.append(stat_info) 
        return stats

After saving the file, restart DataKit and you will be able to see the corresponding metrics on the Guance platform shortly.

Metric List

The sar command can obtain many useful disk metrics. The above script only collect await and svctm. If you need to collect additional metrics, you can modify the script accordingly.

Metric Description Type Unit
await The average time (in milliseconds) for I/O requests issued to the device to be served. This includes the time spent by the requests in queue and the time spent servicing them. float ms
svctm awaitThe average service time (in milliseconds) for I/O requests that were issued to the device. float ms

FAQ

What is the data source on Linux hosts

On Linux hosts, the metrics are parsed and calculated from the /proc/diskstats file; an explanation of each column can be found in procfs-diskstats;

The corresponding relationship between some data source columns and indicators is as follows:

| col04: reads completed successfully | reads | | col05: reads merged | merged_reads | | col06: sectors read | read_bytes = col06 * sector_size; read_bytes/sec = (read_bytes - last(read_bytes))/(time - last(time)) | | col07: time spent reading (ms) | read_time | | col08: writes completed | writes | | col09: writes merged | merged_writes | | col10: sectors written | write_bytes = col10 * sector_size; write_bytes/sec = (write_bytes - last(write_bytes))/(time - last(time)) | | col11: time spent writing (ms) | write_time | | col12: I/Os currently in progress | iops_in_progress | | col13: time spent doing I/Os (ms) | io_time | | col14: weighted time spent doing I/Os (ms) | weighted_io_time |

attention:

  1. Sector size is 512 bytes;
  2. Increment all but read_bytes/sec and write_bytes/sec.

Feedback

Is this page helpful? ×