第5章 Retrieving Metrics

You can retrieve metrics from PCP using the CLI or the web UI interfaces. A number of CLI tools are provided with PCP, which can either output live data, or data from archived sources. The web UI interfaces are provided by the Grafana and Vector web applications. Vector connects directly to the PCP daemon, and can only display live data. Grafana reads from PCP archive files and can display data to up to 1 year old.

5.1. Retrieving Metrics via the CLI

Using the CLI tools provided with PCP, you can retrieve metrics either live, or from an archive file.

5.1.1. Retrieving Live Metrics using CLI

To output metrics on disk partition write instances, enter the following command: 

# pmval -f 1 disk.partitions.write

In this example, PCP converts the number of writes to disk partitions from a counter value, to a rate value. The -f 1 specifies that the value be abbreviated to one decimal place.

Example output

metric:    disk.partitions.write
host:      satellite.example.com
semantics: cumulative counter (converting to rate)
units:     count (converting to count / sec)
samples:   all

                 vda1                  vda2                   sr0
                  0.0                  12.0                   0.0
                  0.0                   1.0                   0.0
                  0.0                   1.0                   0.0
                  0.0                   2.0                   0.0

To monitor system metrics with a two second interval: 

# pmstat -t 2sec

5.1.2. Retrieving Archived Metrics using CLI

You can use the PCP CLI tools to retrieve metrics from an archive file. To do that, add the --archive parameter and specify the archive file.

  • To list all metrics which were enabled when the archive file was created, enter the following command: 

    pminfo --archive archive_file

  • To confirm the host and time period covered by an archive file, enter the following command: 

    # pmdumplog -l archive_file

Examples

  • To list disk writes for each partition, over the time period covered by the archive file: 

    # pmval --archive /var/log/pcp/pmlogger/satellite.example.com/20180816.00.10 \
  -f 1 disk.partitions.write

  • To list disk write operations per partition, with a two second interval, between the time period 14:00 and 14:15: 

    # pmval --archive /var/log/pcp/pmlogger/satellite.example.com/20180816.00.10 \
  -d -t 2sec \
  -f 3 disk.partitions.write \
  -S @14:00 -T @14:15

  • To list average values of all performance metrics, including the time of minimum/maximum value and the actual minimum/maximum value, between the time period 14:00 and 14:30. To output the values in tabular formatting: 

    # pmlogsummary /var/log/pcp/pmlogger/satellite.example.com/20180816.00.10 \
  -HlfiImM \
  -S @14:00 \
  -T @14:30 \
  disk.partitions.write \
  mem.freemem

  • To list system metrics stored in an archive, starting from 14:00. The metrics are displayed in a format similar to the top tool. 

    # pcp --archive /var/log/pcp/pmlogger/satellite.example.com/20180816.00.10 \
  -S @14:00 \
  atop