Skywalking Oap Server crashloop

[linshenfanying]

SkyWalking is deployed on Kubernetes with a separated L1 and L2 cluster setup. However, the L1 nodes frequently restart. What could be the reason? How can we avoid this?

We have utilized SkyWalking for application tracing, and the corresponding architecture is shown in the diagram. A total of 220 services and approximately 1000 instances have been integrated.

The configuration of our OAP Server is 2 CPU cores and 4 GB of memory. We would like to ask for optimization suggestions.

[wu-sheng]

2U4G for 1000 instances? I think OAP is already super powerful.

[linshenfanying]

We are already using Elasticsearch. After changing the configuration of each node to 12C20G, do we need to keep the number of nodes the same?

[wu-sheng]

Can't be sure. Just depend on your payload.

[linshenfanying]

In the SkyWalking cluster, are there any allocation recommendations for the Receiver and Aggregator? In an architecture with Kafka, which one, Receiver or Aggregator, needs more resources?

[wu-sheng]

I don't think you have to set up Receiver and Aggregator if you don't have any specific reason. Just use the mixed role.

[linshenfanying]

There are shared documents on the SkyWalking official website that address the same issues we are encountering, so we will use their approach to solve this problem.