Instead of setting the ingress port on install, make sure to set
the port when the add-on gets loaded (on Supervisor startup and
before installation). This is necessary since the dynamic ingress
ports are not stored as part of the add-on data storage themself
but in the ingress data store. So on every Supervisor start the
port needs to be transferred to the add-on model.
Note that we still need to check the port on add-on update since
the add-on potentially added (dynamic) ingress on update. Same
applies to add-on restore (the restored version might use a dynamic
ingress port).
* Remove race with watchdog during backup, restore and update
* Fix pylint issues and test
* Stop after image pull during update
* Add test for max failed attempts for plugin watchdog
* Backup and restore track progress in job
* Change to stage only updates and fix tests
* Leave HA alone if it wasn't restored
* skip check HA stage message when we don't check
* Change to helper to get current job
* Fix tests
* Mark jobs as internal to skip notifying HA
* Extend container image name validator
The current validator allows certain invalid names (e.g. upper
case), but disallows valid cases (such as ttl.sh/myimage).
Improve the container image validator to support more valid
options and at the same time disallow some of the invalid
options.
Note that this is not a complete/perfect validation still. A much
much more sophisticated regex would be necessary to be 100% accurate.
Also we format the string and replace {machine}/{arch} using Python
format strings. In that regard the image format in Supervisor deviates
from the Docker/OCI container image name format.
* Use an actual invalid image name in config validation
* Add job group execution limit option
* Fix pylint issues
* Assign variable before usage
* Cleanup jobs when done
* Remove isinstance check for performance
* Explicitly raise from None
* Add some more documentation info
* Reduce executor code for docker
* Fix pylint errors and move import/export image
* Fix test and a couple other risky executor calls
* Fix dataclass and return
* Fix test case and add one for corrupt docker
* Add some coverage
* Undo changes to docker manager startup
* Addon startup waits for healthy
* fix import for pylint
* wait_for to 5 in tests
* Adjust tests to simplify async tasks
* Remove wait_boot time from addons.boot tests
* Eliminate async task race conditions in tests
* Add host UTS namespace support for Add-Ons
Using the UTS host namespace is useful when running a mDNS responder
which learns the hostname from the gethostname syscall. This way the
add-on can use the system's hostname without further doing.
* Check host_uts default
* Adjust Security rating if host UTS mode and CAP_ADMIN is set
* Don't add hostname to DNS server if UTS namespace is disabled
* Simplify hostname logic
* Update supervisor/docker/addon.py
Co-authored-by: Mike Degatano <michael.degatano@gmail.com>
---------
Co-authored-by: Mike Degatano <michael.degatano@gmail.com>
* Support for Docker manifests base images add-on build
* Set platform for build and tests
* Remove empty test
Co-authored-by: Mike Degatano <michael.degatano@gmail.com>
* Add update freeze option
* Freeze to auto update and plugin condition
* Add tests
* Add supervisor_version evaluation
* OS updates require supervisor up to date
* Run version check during startup
* Docker events based watchdog
* Separate monitor from DockerAPI since it needs coresys
* Move monitor into dockerAPI
* Fix properties on coresys
* Add watchdog tests
* Added tests
* pylint issue
* Current state failures test
* Thread-safe event processing
* Use labels property
* Return list of possible data disk targets
* fix path
* fix tests
* Add test
* Fix tests
* Add tests
* Add more tests
* Remove debug
* Address comments
* more clear
Stefan Agner
J. Nick Koston
Mike Degatano
Pascal Vizeli
Franck Nijhof
Joakim Sørensen