K8s ecological weekly | etcd v3.4.0 brings many updates


The content of k8s ecological weekly mainly includes some recommended information about k8s ecology that I have been exposed to every week. Welcome to the k8s ecology column.

Helm 3 beta 2 release

Helm 3 has been in continuous improvement since alpha. Finally, beta2 has been released. According to the current development progress of the community, there is still a lot of hope to release the official version this year.

If you are interested, you can try it first to avoid discomfort when upgrading later.

Coredns v1.6.3 release

  • federationWill be completely abandoned in 1.7.0;
  • Two new plug-insclouddnsandsignAmong themclouddnsAs the name implies, it is designed for the cloud environment. Now it supports the zone data provided by GCP (Google cloud platform) cloud DNS. In fact, it obtains these information through the Google cloud API. If you are not using GCP cloud DNS, this plug-in should not be used at present.signThe plug-in uses nsec signature for zone according to RFC 6781, but it should be noted that the signature is time-effective. If the expiration time is reached, the zone information will become bad status (RFC 4035), so if you want to use this plug-in, please know clearly what you need to do and why you use it;
  • fileThe plug-in fixes some memory leaks.

In addition to the above, I would like to mention a little more about the newazurePlug in, it’s actuallyclouddnsWhat we do is similar, just to get records from azure. In addition, starting from v1.6.1reloadThe plug-in can correctly check the changes of the files imported in corefile.

For more information about this release, refer to release note

Strimzi becomes CNCF sandbox project

Strimzi is a project dedicated to running Apache Kafka on kubernetes, but before it applied to become CNCF sandbox project this time, another direction is to let Apache Kafka run on openshift. In order to become the sandbox project of CNCF, most of the content you can see on the website has been rewritten to run on kubernetes.

So if you are using openshift, you can also try this project. Attached is a structure chart

K8s ecological weekly | etcd v3.4.0 brings many updates

Etcd v3.4.0 officially released

Etcd v3.4.0 has been released recently, which includes a large number of updates.

  • Add raft learner: direct viewing of design documents is recommended
  • Back end concurrent read: with this improvement, write throughput can be increased by 70%, p99 write latency reduced by 90%, and related PR;
  • The implementation of load balancing of client V3 is rewritten by using grpc balancer API: in this process, of course, grpc dependency is upgraded to v1.23.0, and the problem of etcd encountered in kubernetes is also corrected. For details, please refer to K / K ﹐ 72102.
  • ETCDCTL_API=3 etcdctlandetcd --enable-v2=falseBecomes the default configuration;
  • Added a lot of metrics information;
  • Another major change is that go 1.12 + is now required, and dependency management alsoglideChange to go module;

There are many updates this time. It is recommended to check the documents in detail before upgrading.

For more information about this release, refer to release note

Upstream development progress

  • The coredns version used by kubeadm by default has been updated to v1.6.2. If you have been reading the k8s weekly ecological report, the relevant changes need to be noted are basically discussed. However, to mention a little more, the current deployment of coredns is approved byreadyPlug in is used for readiness check, but there are problems before; in additionproxyThe plug-in has been discarded, so if you want to upgrade the cluster, you need to pay attention to whether corefile needs to be adjusted.
  • Kubeadm’s kernel verification rules are changed to support all 3.10 + kernels.

You can subscribe to my article public number [moelove] through the following QR code, and reply k8s in the background of the public number to join the technical circle.

K8s ecological weekly | etcd v3.4.0 brings many updates