Health-Check

An health-check is specific LAVA jobs that is automatically and regularly scheduled to check the health of DUT.

If for any reason the job fails, the DUT health will be set to Bad (see device health). The device will not be used anymore by the scheduler until an admin set the health to either unknown or good.

Reports

Health-check reports are available at http://localhost/scheduler/reports. The report shows health-check failures and track the general health of the devices.

There is also a table providing information on the fails checks at http://localhost/scheduler/reports/failures?health-checks=1.

Recommendations

Golden image

In order to provide constant results, we advice to only use Golden image for health-checks.

The infrastructure

As health-checks are normal LAVA jobs, an health-check can test any part of the infrastructure that is normally used by a LAVA job.

For instance, we recommend to test:

• bootloader methods
• attached devices (hard-drive, probes, ...)
• network access (local or remote servers)

lava-test-raise

In order to fail a job during the test, call lava-test-raise <message>. This will return immediately and set the job health to Incomplete.

Configuration file

The health-checks are stored on the server in /etc/lava-server/dispatcher-config/health-checks/<name>.yaml

Admin could update device-type health-check using lavacli:

lavacli device-types health-check set qemu qemu.jinja2

filename

In order to compute the health-check of a DUT, LAVA will look in the device dictionary for the {% extends device-type.jinja2 %} line. The health-check filename is device-type.yaml.