Releases: rancher/system-upgrade-controller
Releases · rancher/system-upgrade-controller
v0.6.2
v0.6.1
Features and Enhancements
- Support DNS Policy
ClusterFirstWithHostNet
#74 - Support Drain of Pods with Disruption Budget for
kubectl
v1.18 or newer #79
This leverages the new-as-of-v1.18--disable-eviction
flag ofkubectl
.
If yourkubectl
image is not v1.18 or newer your plans will fail to apply with${plan}.spec.drain.disableEviction == true
- Support SkipWaitForDeleteTimeout for
kubectl
v1.18 or newer #89
This leverages the new-as-of-v1.18--skip-wait-for-delete-timeout
flag ofkubectl
.
If yourkubectl
image is not v1.18 or newer your plans will fail to apply with${plan}.spec.drain.skipWaitForDeleteTimeout > 0
- Support Plan Validation #80
Attempting to push a plan to the apiserver that is missing.spec.upgrade
should now be rejected. This avoids a panic in the controller and encourages sanity. - Support Common Tolerations in Deployment #84
Bug Fixes
- Prevent Over-Allocation of Jobs #83
Labelling multiple nodes in one invocation ofkubectl
would sometimes result in more jobs than the specified concurrency. The controller now checks for this and deletes jobs for nodes that are not on the applying list.
Other
- Avoid kustomize v3.5.5 at build-time #75
currently pinned to v3.5.4
v0.6.0
Features and Enhancements
- Support DNS Policy
ClusterFirstWithHostNet
#74 - Support Drain of Pods with Disruption Budget for
kubectl
v1.18 or newer #79
This leverages the new-as-of-v1.18--disable-eviction
flag ofkubectl
.
If yourkubectl
image is not v1.18 or newer your plans will fail to apply with${plan}.spec.drain.disableEviction == true
- Support SkipWaitForDeleteTimeout for
kubectl
v1.18 or newer #89
This leverages the new-as-of-v1.18--skip-wait-for-delete-timeout
flag ofkubectl
.
If yourkubectl
image is not v1.18 or newer your plans will fail to apply with${plan}.spec.drain.skipWaitForDeleteTimeout > 0
- Support Plan Validation #80
Attempting to push a plan to the apiserver that is missing.spec.upgrade
should now be rejected. This avoids a panic in the controller and encourages sanity. - Support Common Tolerations in Deployment #84
Other
- Avoid kustomize v3.5.5 at build-time #75
currently pinned to v3.5.4
v0.5.0
- Support Kubernetes 1.18 #53
- Support Plan Latest Version as Prepare Container Tag #61
Also, do not override tags provided in either theprepare
orupgrade
container spec. - Fix Panic On Plan Delete #65
- Die, Flappy Bird, Die #67
Addresses and issue, #58, resulting ink3os-latest
jobs form prior upgrades causing intermittent unavailability. Stale jobs will be deleted with prejudice. - Support Plan Tolerations #68
v0.5.0-m1
v0.4.0
v0.4.0-rc2
v0.4.0-rc1
v0.3.1
v0.3.0
- Support Upgrade Job Running with
HostNetwork=true
#21 - Ginkgo/Gomega BDD-Style Unit Tests #20
- Sonobuoy Plugin for Running E2E Tests #26
- Support Munging of SemVers with Metadata #29
- Fix Clashing Job/Pod Names #33
- Fix Replace/Wait Errors #30
- Fix Extra Job Generation #39
- Example Plan for Ubuntu Bionic #39