“K8S Ecology Weekly” mainly contains some recommended weekly information related to K8S ecology that I have come into contact with. Welcome to subscribe zhihu column “K8S Ecology”.
The Helm V3.2.1 version is released
This week Helm released v3.2.1, the first patch release in the V3.2 series. Some things to watch for this time:
-
#7959 fixed a bug that had been present at Helm V3 since version 3.0-RC. For details, see #6899, but — repurse-values is not used by many people.
-
#7653 Changed the helm upgrade behavior to allow helm Upgrade to be redeployed on a failed Release without removing the old Release first; (Much more convenient!)
In fact, Helm also released V2.16.7 this week. Although Helm 2 has entered the maintenance period, we have to say that the maintenance team is still very reliable and has been continuously working on patch.
Continue to recommend upgrading to Helm V3.
Those interested in this release can check out the complete ReleaseNote.
Rancher v2.4.3 release
Rancher V2.4 is a big release, and this release includes a number of updates and fixes. Take a look!
- For RHEL/CentOS node pool users, the default Docker storage driver has been changed
overlay2
Special attention should be paid here. The originaldevicemapper
Storage drivers have been marked as abandoned in Docker, which has been explained in previous weekly reports and will not be repeated here. - The Feature Flags option for the experimental new Feature, which you can specify with the startup parameter or select directly from the UI, is available in the documentation at rancher.com/docs/ranche… ;
- #21361 Fixed a Rancher memory leak that could cause OOM issues
- #26061 Fixed slow login when there are lots of LDAP groups
There are also some other fixes and improvements that interested friends can refer to in ReleaseNote
Progress in the upstream
- #88915 Fixed an issue in Kubelet Image Manager that could cause static Pod workers to stop working. This issue has been documented since 1.10, but I haven’t encountered it in production yet. Upgrade as soon as possible though;
- #89660 Kubectl output format will provide an option to output JSONPATH as JSON,
Example output is as follows:
old:
$ /old/kubectl get pod -o=jsonpath='{ ..'ip' }'New/Unchanged: $kubectl get pod -o=jsonpath='{ ..'ip' }'10.42.0.50 10.42.0.49 10.42.0.48 new: $kubectl get pod -o=jsonpath-as-json='{ ..'ip' }'
[
"10.42.0.50"."10.42.0.49"."10.42.0.48"
]
Copy the code
- # 88723
kubectl taint
Become more flexible, no longer need to specify all resource names;
Please feel free to subscribe to my official account [MoeLove]