Device Capabilities Support

Device Capabilities Support

Device Capabilities include ways for individual users and groups to get more information on LAVA devices, control who can submit jobs to particular devices and manage device transitions without needing full administrator access.

The user or group with physical access to the device is also stored, this is useful when someone needs to interact with the device directly rather than through network, serial or LAVA connections.

The Device Owner is a dedicated field in the admin interface which displays the email address of the owner or the name of the group.

New device owners can only be created by the lab administrators. Equally, assigning or transferring ownership of any device can only be done by the lab administrators.

LAVA also identifies certain Device capabilities and information from the health check jobs run on the device.

Device owner abilities

  • add free text comments to a device status transition
  • initiate a device status transition on an owned device
  • cancel any “current” job running on the device.
  • restrict submissions to the device to just the owner for a period of time - (when unrestricted, devices remain in the general usage pool.) MultiNode submissions by the owner can use restricted and pool devices. Restricted devices are not included in the count of devices available during job submission by other users.
  • Edit device-specific timeouts for stages within dispatcher processes. This allows the hardcoded values to be reset to general usage defaults and devices which need exceptions to specify such exceptions individually.
  • Edit device description - could include particular hardware support on this device or particular hardware constraints. Could also include details of why the device is currently restricted, beyond the short message sometimes added to the transition status change.
  • Change the priority for any submitted job which reserves an “owned” device. Priority can only be changed once an owned device has been reserved for the job and before the job starts running or an error will be returned.
  • annotate individual job failures
  • skip health check this time only when putting a device online. This omission is recorded in the state transition log for the device.
  • force a health check without needing to take the device offline and then back online.