Replies: 4 comments 1 reply
-
|
I can appreciate that a default route is pretty common scenario, but it doesn't inherently mean there is connectivity either. Flannel for example has a similar approach, and they had to offer I am happy to produce a PR to make this more configurable, if there is agreement that this can be revised? |
Beta Was this translation helpful? Give feedback.
-
|
We probably need to expose network probes in the machine configuration to make it easier to configure, and in general we can make connectivity check configurable. This shouldn't be painful. |
Beta Was this translation helpful? Give feedback.
-
|
Here is a patch to expose the probes in machine configuration: #12606 |
Beta Was this translation helpful? Give feedback.
-
|
This could also work for us (ability to declare a blackhole as default gateway): #12608 |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Context
I'm running Talos on OpenNebula (but the issue exists also for other platforms I suppose) in a network environment where we don't have a default gateway route (0.0.0.0/0). The network configuration is managed through the machine config, and connectivity goes through a corporate proxy.
Issue
Talos seems to be stuck waiting for connectivity, and from looking at the code (internal/app/machined/pkg/controllers/network/status.go), I can see that
ConnectivityReadyrequires either:Since we don't have a default route, I'm trying to understand how to use network probes.
What I Found
I discovered that there's a probe mechanism in Talos that can validate connectivity by checking TCP endpoints. Looking at the code, I can see:
PlatformConfigSpechas a Probes fieldQuestions
talos.platform=metalon OpenNebula? Would this allow me to use META partition for probe configuration and machine configuration, or would it break other things?I'd appreciate any guidance on how to handle this scenario. Thanks!
Beta Was this translation helpful? Give feedback.
All reactions