Red Hat Bugzilla – Bug 1316302
[RFE] Diagnostics pod that can be deployed to test a new nodes and successful install.
Last modified: 2016-10-31 07:30:32 EDT
Description of problem:
- Requesting a quickstart template that would deploy a pod that runs on all nodes or a single node that would test for a successful install of the cluster or addition of a node.
- Ensures pods have access to external network
- Makes sure that DNS is configured
- Checks if services are started and interfaces configured
- Runs `openshift ex diagnostics`
This would be an enhancement of the existing diagnostic DiagnosticPod. It doesn't run on all nodes or check external connectivity yet.
BTW tell people to use "oadm diagnostics" - it's graduated from experimental status.
To test external connectivity requires a viable external target to attempt to connect to, as well as awareness of any proxy params. Do you have an opinion on what external target would be acceptable? "www.google.com" perhaps. The other complication is that the customer may not *want* external connectivity, and I don't know of a good way to detect that intent. So this would probably be a warning if it failed.
What do you mean by services started and interfaces configured? Which services and which interfaces? Are you thinking this test pod would have privileged access to everything on the node and basically run a local "oadm diagnostics" on every node?
The latest command checks DNS and network. I'm fine with this being closed