Test checks Plugins

Common Keys

The following keys are accepted by all plugins of the test-checks step.

enabled: true|false

Whether the check is enabled or not.

Default: true

avc

Check for SELinux AVC denials raised during the test.

The check collects SELinux AVC denials from the audit log, gathers details about them, and together with versions of the selinux-policy and related packages stores them in a report file after the test.

check:
  - name: avc

Note

To work correctly, the check requires SELinux to be enabled on the guest, and auditd must be running. Without SELinux, the check will turn into no-op, reporting skip result, and without auditd, the check will discover no AVC denials, reporting pass.

If the test manipulates auditd or SELinux in general, the check may report unexpected results.

Added in version 1.28.

dmesg

Save the content of kernel ring buffer (aka “console”) into a file.

The check saves one file before the test, and then again when test finishes.

check:
  - how: dmesg

Check will identify patterns that signal kernel crashes and core dumps, and when detected, it will report as failed result. It is possible to define custom patterns:

check:
  - how: dmesg
    failure-pattern:
      # These are default patterns
      - 'Call Trace:
      - '\ssegfault\s'

      # More patterns to look for
      - '\[Firmware Bug\]'

Added in version 1.28.

Changed in version 1.33: failure-pattern has been added.

Configuration

See also Common Keys accepted by the plugin.

failure-pattern: PATTERN

List of regular expressions to look for in dmesg output. If any of patterns is found, dmesg check will report fail result.

Default: Call Trace:, \ssegfault\s

watchdog

Warning

Please, be aware that the documentation below is a work in progress. We are working on fixing it, adding missing bits and generally making it better. Also, it was originally used for command line help only, therefore the formatting is often suboptional.

Take various actions when guest becomes unresponsive.

Watchdog runs selected probes every now and then, and when a given number of probes fail, watchdog would run one or more of the predefined actions.

Check comes with two probes, “ping” and “SSH ping”, and single action, “reboot”.

“ping” uses the classic ICMP echo to check whether the guest is still up and running,
“SSH ping” tries to establish SSH connection,
“reboot” action issues a hard reboot of the guest.

Warning

Be aware that this feature may be limited depending on how the guest was provisioned. See Hard reboot.

Each probe has a “budget” of allowed failures, and when it runs out, the action is taken. A successful probe replenishes its budget to the original level.

Multiple probes can be enabled at the same time, for the action to happen it’s enough if just one of them runs out of its budget.

check:
  - how: watchdog
    ping: true
    reboot: true

check:
  - how: watchdog

    # Use only SSH ping.
    ping: false
    ssh-ping: true

    # Try every 5 minutes, allow 7 failed attempts, and reboot
    # the guest when we run out of attempts.
    interval: 300
    reboot: true
    ssh-ping-threshold: 7

Added in version 1.32.

Configuration

See also Common Keys accepted by the plugin.

interval:

How often should the watchdog run, in seconds.

Default: 60

ping: true|false

If enabled, watchdog would probe guest’s responsiveness with ICMP packets.

Default: false

ping-packets:

How many ICMP packates to send as one probe.

Default: 1

ping-threshold:

How many failed ping probes before taking any further action.

Default: 10

reboot: true|false

If enabled, watchdog would reboot the guest after enough failed probes.

Default: false

ssh-ping: true|false

If enabled, watchdog would probe guest’s responsiveness by connecting to its SSH port.

Default: false

ssh-ping-threshold:

How many failed SSH connections before taking any further action.

Default: 10