The author: Source Third, the Pro Stone, Zhang Lei, Mo source
Industry News
1. K8s 1.16 to a series of abandoned old API version
The impact was involved NetworkPolicy, PodSecurityPolicy, DaemonSet, Deployment, StatefulSet, ReplicaSet and Ingress. Please K8s users and developers concerned, all the relevant API migration were as follows:
- NetworkPolicy: no longer in use extensions / v1beta1 in the v1.16;
- Migrating to networking.k8s.io/v1 API, available since v1.8, and update inventory data can be obtained through the new API.
- PodSecurityPolicy: no longer in use extensions v1.16 / v1beta1;
- Migrating to policy / v1beta1 API, since v1.10 is available, and update inventory data can be obtained through the new API.
- Deamon Set, Deployment, StatefulSet and ReplicaSet: v1.16 from start through no extension / v1beta1, apps / v1beta1, or apps / v1beta2 provided;
- Migrating to apps / v1 API, since v1.9 is already available, and update inventory data can be obtained through the new API.
- Ingress: From the beginning v1.18 longer available through extensions / v1beta1;
- Migrating to networking.k8s.io/v1 API, stock data can get more information and updates through the new API: https://kubernetes.io/blog/2019/07/18/api-deprecations-in-1-16/
2. Prometheus continued high-profile, a number of manufacturers introduced a managed or cloud integration services
As another successful project under CNCF, Prometheus has been to integrate with Azure Monitor on Microsoft Azure, is now in preview stage . Last month, Ali cloud launch a hosted version of Prometheus monitoring products to support the black and white of the installation exporter, monitoring the market, open source components out of the box fully compatible without basic operation and maintenance capabilities for free. In addition, Ali cloud also launched the open source enhancement of Prometheus solutions , the enhancements made in acquiring index richness, accuracy and other indicators collected to support the use Ali cloud TSDB timing database as persistent and highly available data can be simple and convenient one-Click management installed by Helm Chart.
Significant progress upstream
Kubernetes design enhancements proposed (KEP)
- IPv6 support into the Beta stage ;
- Cloud Provider Label prepare GA current cloud provider label is beta, remove and modify the plan.
Knative project
- Plans to release August 6 Serving 0.8, the relevant issue is mainly to improve the availability and stability;
- Enhance the ease of use of consumer event directly from the source, determine the extension Knative CLI scenes and events need to be modified to use the model.
Open source projects recommended
head
Python for a user Kubernetes Operator Framework. It provides a concise set of primitives, such that a user can use a simple Python code to achieve a fast Operator, and these primitives masked Operator technical details, focusing on the logical operation and maintenance inside the Operator.
This week Recommended Reading
《Best Practices: Benchmarking Service Mesh Performance》
This paper introduces the performance of the Service Mesh (Istio) of best practices for Benchamark.
《451 Research 的 Cloud Price Index》
The introduction of third party business surveys, provided its analysis and insight for public and private cloud prices in different regions of the world.
《Cloud-Native CI/CD with OpenShift Pipelines》
It describes the release of the OpenShift Pipelines in OpenShift 4.1 Developer Preview (developer preview), OpenShift Pipelines which is OpenShift integration practices of Tekton project.
《Avoid time-of-measurement bias with Prometheus》
We currently have many tools (such as Prometheus) to monitor the performance of one of our Server, but in many cases, a Server service is asynchronously provided by the worker behind a lot of.
Often it happens in practice is: although we, but we still do not know from a variety of asynchronous service worker who Metrics exactly what to do, and this often leads to us (although a pile of hand tools) can not quickly locate problem. This blog via a classic case describes the pain points and practical approach which, at the same time introduced the open source tools: https://github.com/lawrencejones/prometheus-client-tracer-ruby