v1.7.0

Official v1.7.0 release of Cyclops!

🗒️ Changelog

• Perform health checks before and during cycling to ensure workloads are stable throughout
• Stability fixes

🐳 Container Image

ghcr.io/atlassian-labs/cyclops:v1.7.0

v1.6.0

Official v1.6.0 release of Cyclops!

🗒️ Changelog

• Force delete pods off the node if they have been unhealthy for a while when cycling

🐳 Container Image

ghcr.io/atlassian-labs/cyclops:v1.6.0

v1.5.0

Official v1.5.0 release of Cyclops!

🗒️ Changelog

• Cyclop binary can use the flag --default-cns-cycling-expiry=<time.Duration-format> to change the default timeout duration
• Cyclop CLI can use the flag cycling-timeout=<time.Duration-format> to customise the timeout for the generated CNR
• New updates to NodeGroup, CycleNodeRequest, and CycleNodeStatus CRDs to ultilise the new functionalities

🐳 Container Image

ghcr.io/atlassian-labs/cyclops:v1.5.0

v1.4.0

Official v1.4.0 release of Cyclops!

🗒️ Changelog

• Fix an issue where cyclop-observer scaling avoidance do not detect cluster-autoscaler scale up event

🐳 Container Image

ghcr.io/atlassian-labs/cyclops:v1.4.0

v1.3.0

Official v1.2.0 release of Cyclops!

🗒️ Changelog

• Bump go version, dependencies update, remove vulnerable dependencies

🐳 Container Image

ghcr.io/atlassian-labs/cyclops:v1.3.0

v1.2.0

Official v1.2.0 release of Cyclops!

🗒️ Changelog

• Fixed race condition between ClusterAutoscaler and Cyclops
  If prometheusAddress argument is specified, Cyclops will look for cluster_autoscaler_last_activity{activity='scaleUp'} events in Prometheus. If any are found within the nodeStartupTime (default 2min), Cyclops will wait before creating a new CNR.

🐳 Container Image

ghcr.io/atlassian-labs/cyclops:v1.2.0

v1.1.1

Official v1.1.1 release of Cyclops!

🗒️ Changelog

• Fixed flip-flopping behaviour when a CycleNodeStatus object times out due to a pod not getting drained.
• Fixed issue where under certain circumstances, Cyclops repeatedly transition to Failed.
• Added a check to ensure that the CycleNodeRequest object name is also a valid label name and not, the CycleNodeRequest object is transitioned to Failed before any cycling takes place.
• Fixed visual bug with the Nodegroups section in the slack status message.

🐳 Container Image

ghcr.io/atlassian-labs/cyclops:v1.1.1

v1.1.0

Official v1.1.0 release of Cyclops!

🗒️ Changelog

Known issue: this release is know to cause Cyclops to transition from Initialised to Successful immediately after detaching an instance. We currently recommend using release v1.0.0.

• Added a new field to the CycleNodeRequest CRD named NumNodesCycled which keeps track of the number of nodes cycled. Please make sure to update the CycleNodeRequest CRD in your cluster before running Cyclops or else you will encounter unexpected behaviour.
• Changed the total nodes count in the notifications to account for a CycleNodeRequest to cycle an entire NodeGroup by omitting the nodeNames field.
• Improved the Progress field in the main status notification to count the numbers of nodes successfully cycled instead of nodes selected for cycling.
• Fixed the Nodegroup field in the main status notification to account for multiple NodeGroups listed in the NodeGroupList field in the CycleNodeRequest.

🐳 Container Image

ghcr.io/atlassian-labs/cyclops:v1.1.0

v1.0.0

Official v1.0.0 release of Cyclops!

🗒️ Changelog

• Promotes v0.0.4 to v1.0.0

🐳 Container Image

ghcr.io/atlassian-labs/cyclops:v1.0.0

v0.0.4

First tagged release of Cyclops

🗒️ Changelog

• Pushed image to Github Container Registry

🐳 Container Image

ghcr.io/atlassian-labs/cyclops:v0.0.4