add insertable = false on same places where hibernate validatire requires
the purpose is that code become more less eclipselink bound and to be closer to potentially hibernate supported
Signed-off-by: Avgustin Marinov <Avgustin.Marinov@bosch.com>
Now it is not part of the ddi and dmf starters - only mgmt (where could be excluded on packaging)
Signed-off-by: Avgustin Marinov <Avgustin.Marinov@bosch.com>
* it's not needed for dmf
* also made optional when not needed (e.g. some JpaArtifactManagement)
_release_notes_
Signed-off-by: Avgustin Marinov <Avgustin.Marinov@bosch.com>
* added methods to unassign by multiple ds
* deprecated toggle assigments - too complex to undestand
* deprecated unassign (management) of single ds - in favour of methods with multiple ds
Signed-off-by: Marinov Avgustin <Avgustin.Marinov@bosch.com>
* added methods to unassign by multiple controller ids
* deprecated toggle assigments - too complex to undestand
* deprecated unassign (management) of single controller id - in favour of methods with controller ids
Signed-off-by: Marinov Avgustin <Avgustin.Marinov@bosch.com>
when in StartNextGroupRolloutGroupSuccessAction#startNextGroup:
1. start all scheduled actions
2. if started are > 0 -> RUNNING, otherwise -> FINISHED (if not dynamic rollout)
what could possibly happen is that at same time:
* because of a success condition met the JpaRolloutsExecutor triggers start the group
* user triggers start of the next group (via RolloutsManagement#triggerNextGroup)
then it could:
* the 'first' one succeeds to start next group
* the second attempts to start it (JpaRolloutsExecutor found the previous had met the success condition or trigger next found it SCHEDULED and next to run)
* the second finds no scheduled actions (just running) and decides there are no actions. So, it assumes (wrongly) no actions in group - and set it as FINISHED
This way we could have FINISHED group with still running actions
Signed-off-by: Marinov Avgustin <Avgustin.Marinov@bosch.com>
* Quota must be checked with conditions over incoming status, not current persisted in db
* Fix Download_Only case where DOWNLOADED is threated as 'final'.
Fix ci build tests.
* Review findings
Adding a method with:
* optimized payload - just controller ids
* no response payload - not needed for that use-case
* targeting - thousands of targets tagged at once
Signed-off-by: Marinov Avgustin <Avgustin.Marinov@bosch.com>
At the moment such unused are removed for "tag!=1 and tag==2" but not for "tag==2 and tag!=1"
Signed-off-by: Marinov Avgustin <Avgustin.Marinov@bosch.com>
* Fix RSQL filter for no target tag and OR
* add test for such filter
* Clean up the code
keeps the legacy Rsql Visitor which could be used with hawkbit.rsql.legacyRsqlVisitor=true
---------
Signed-off-by: Marinov Avgustin <Avgustin.Marinov@bosch.com>
* involve targets into dynamic rollouts eagerly - doesn't wait for dynamic group to become running in order to involve devices
* adds trottling for involving targes into dynamic groups
* small style refactoring
Signed-off-by: Marinov Avgustin <Avgustin.Marinov@bosch.com
Signed-off-by: Marinov Avgustin <Avgustin.Marinov@bosch.com
* feat: allow a target to set offline assigned distribution set
Signed-off-by: Florian Bezannier <florian.bezannier@hotmail.fr>
* refacto: apply @avgustinmm recommendation
* docs: Mark update offline API as experimental
---------
Signed-off-by: Florian Bezannier <florian.bezannier@hotmail.fr>
