Move from: bobsql.com
If you are new to Kubernetes (K8’s), like I am, you quickly learn that the servicing node’s journal holds a wealth of information.
In the lab I was encountering evictions due to memory, disk space and other resource limitations. My collection scripts were connecting to the pods and not the nodes. This approach lacks valuable, lineage information needed to fully troubleshoot issues.
For example, my master-0, SQL Server, pod was getting evicted and restarted. I used SSH to connect to the master-0 pod.
Note: the debug information and logs located on the pod