Tests may specify attributes via the Attr
field in testing.Test. Attributes are free-form strings, but this document enumerates well-known attributes with established meanings.
The following attributes may be added to control how tests are run and how their results are interpreted.
Zero or more group:*
attributes can be set to a test to assign it to groups. A group is a collection of tests having similar purposes or characteristics, often (but not necessarily) scheduled together. In automated testing, tests to run are usually selected by groups. If a test belongs to no group, it has no attribute group:*
assigned, that means it will be disabled in automated testing.
Some groups define extra attributes which annotate tests with extra information. They can be set if a test belongs to corresponding groups.
Below is the list of most popular groups and their associated extra attributes:
group:mainline
- The default group for functional tests. Tests having this attribute are called mainline tests. A mainline test falls under exactly one of the two categories: informational if it has informational
attribute; otherwise it is critical. Failures in critical tests justify rejecting or reverting the responsible change, while failures in informational tests are ignored. All informational mainline tests are supposed to be promoted to critical tests. Details on adding tests can be found in this tast-add-test document (googlers only).group:criticalstaging
- This group will be used to indicate a test is intended on going into “mainline” critical testing. This group will be run on all boards/models; on ToT only. Tests can only remain in this group long enough to gather signal (10 consecutive builds), after which the owner must promote them into mainline only, or back into informational. If no owner action is taken after a 4 builds grace period, they will be moved into informational.informational
: This group indicate a test is not critical.group:hw_agnostic
- A test that is not dependent on physical HW and can be run on x86 VMs.vm_stable
: This indicates that a test is stable on VMs. It is intended for mainline informational tests that have yet to be stabilized on all boards/models, but is safe and stable on VMs. This enables such tests to be scheduled and tested on VM only.group:appcompat
- A group of ARC app compatibility tests. Below are its sub-attribute:appcompat_release
: A group of ARC app compatibility tests for release testing.appcompat_smoke
: A group of ARC app compatibility tests for smoke testing.appcompat_top_apps
: A group of ARC app compatibility tests for top apps testing.group:arc
- A group of ARC Functional tests. Below are its sub-attribute:arc_playstore
: A group of ARC tests which verify playstore functionality after updating the playstore to nightly build.arc_core
: A group of ARC tests which verify ARC Core functionality.arc_chromeos_vm
: A group of ARC tests which run ARC functionality on ChromeOS VM.group:crosbolt
- Test failures are ignored and the test's performance data are uploaded to crosbolt. When you add this attribute, you also need to add one of crosbolt_perbuild
, crosbolt_nightly
or crosbolt_weekly
.group:wificell
- Tests that will run on the wificell fixture. Typically those tests require special hardware (APs, ...) only available in those fixtures. Some WiFi-specific functional tests that do not technically require a wificell are also made part of this group for consistency to simplify the validation of the whole WiFi stack. Its sub-attributes can be classified into two types:wificell_func
: verify basic WiFi functionalities nightly.wificell_suspend
: verify basic WiFi behavior related to suspend/resume nightly.wificell_cq
: Similar to wificell_func, but triggered by CLs that touch specific code paths.wificell_perf
: measure WiFi performance.wificell_stress
: Stress test the WiFi functionalities.wificell_mtbf
: measure Mean Time Between Failures (MTBF).wificell_unstable
is present, the test is yet to be verified as stable; otherwise, the test is stable.group:wificell_roam
- Tests that depends on grover (Googlers only) fixture to run. Subattributes that specify role of the test (required):wificell_roam_func
: verify basic WiFi roaming functionalities.wificell_roam_perf
: measure WiFi roaming performance.group:labqual
- Tests that must pass for devices to go to a low-touch lab.group:mtp
- Tests that will run on the mtpWithAndroid
fixture. Typically those tests require special hardware (Android Phone Connected to DUT) and setup available in the fixture.group:cuj
- A group of CUJ tests. Tests having this attribute will have their performance data collected and sent to TPS dashboard. Below are its sub-attributes:cuj_experimental
: A group of CUJ tests which are experimental and only run on a selected subset of models.cuj_weekly
: A group of CUJ tests that run weekly. See attr.go for the full list of valid attributes.Several attributes are also added automatically:
bundle:<bundle>
- Test's bundle, e.g. cros
(automatically added).dep:<dependency>
- Test software dependency (automatically added).name:<category.Test>
- Test's full name (automatically added).See the Running Tests document for information about using attributes to select which tests to run.