As far as I can tell, based on the implementation in MockProvider
NotifyPods is called with the mutated pod. This allows us to
take a copy of the Pod object in NotifyPods, and make it so
(eventually) we don't need to do a callback to GetPodStatus.
This makes sure the update function works correctly after the pod
is running if the podspec is changed. Upon writing the test, I realized
we were accessing the variables outside of the goroutine that the
workers with tests were running in, and we had no locks. Therefore,
I converted all of those numbers to use atomics.
This seems to avoid a race conditions where at pod informer
startup time, the reactor doesn't properly get setup.
It also refactors the root command example to start up
the informers after everything is wired up.
Otherwise the variable is evaluated before it's even set (assuming it's
not set from the CLI).
skaffold also requires bin/e2e/virtual-kubelet to work on it's own.
Vendoring seems to be problematic with go modules.
In addition to causing issues with transitive dependencies and being
problematic when trying to pull in the node-cli code, it also seems to
be pulling different versions than what is computed in go.mod.
This can be seen by comparing commits from the computed version in
go.mod with very small diffs (like a single Godep change in k8s) but massive
vendoring changes.
Basically I have very little confidence that this is even working.
This should be better in go1.13, but this is not released yet.
* Fix error handling for delete pod
- Error handling was looking for a k8s error from the provider, but
providers should be using errdefs.
- Error handling was returning early if pod was not found and deleting
from k8s in all other cases.
* Don't run unit tests twice
* Pointed links to the README.md file in the base of each provider repo
* Removed some provider specific content from the main repos README.
Signed-off-by: Graham Hayes <gr@ham.ie>
This eliminates all the one-off calls to for specific things like
`Capacity` and other things.
Instead we configure defaults in the CLI and then pass the full node
object to the provider to change as needed.
This makes sure the provider can change whatever they need to without
having to add tons of methods to the provider interface.
This allows `--cluster-domain` to be passed to virtual kubelet like a
traditional kublet, and use this to generate search-domains for
`/etc/resolv.conf`
* Set default `cluster-domain` to `cluster-local` to match current kubelet
Related: #641
Signed-off-by: Graham Hayes <gr@ham.ie>
This includes updates to CI config, vendor files, etc.
I've hard-coded the k8s depedency at 1.13.4 to keep it inline with what
we currently have and to make sure a another run of `go mod tidy`
doesn't accidentally update it to an unexpected version.
Thanks to hectorj2f for carrying this along.
