Several places in the code used a 5-second retry loop to wait on
Runtime.Core to be set. This caused a race condition where OnChange
handlers could be added after the Wrangler shared informers were already
started. When this happened, the handlers were never called because the
shared informers they relied upon were not started.
Fix that by requiring anything that waits on Runtime.Core to run from a
cluster controller startup hook that is guaranteed to be called before
the shared informers are started, instead of just firing it off in a
goroutine that retries until it is set.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
We need to send the full chain in order for cross-signing to work
properly during switchover to a new root.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Turns out etcd-only nodes were never running **any** of the controllers,
so allowing multiple controllers didn't really fix things.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Prevents errors when starting with fail-closed webhooks
Also, use panic instead of Fatalf so that the CloudControllerManager rescue can handle the error
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Allow bootstrapping with kubeadm bootstrap token strings or existing
Kubelet certs. This allows agents to join the cluster using kubeadm
bootstrap tokens, as created with the `k3s token create` command.
When the token expires or is deleted, agents can successfully restart by
authenticating with their kubelet certificate via node authentication.
If the token is gone and the node is deleted from the cluster, node auth
will fail and they will be prevented from rejoining the cluster until
provided with a valid token.
Servers still must be bootstrapped with the static cluster token, as
they will need to know it to decrypt the bootstrap data.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
* General cleanup of test-helpers functions to address CI failures
* Install awscli in test image
* Log containerd output to file even when running with --debug
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
* Add EncryptSecrets to Critical Control Args
* use deep comparison to extract differences
Signed-off-by: Derek Nola <derek.nola@suse.com>
Signed-off-by: Derek Nola <derek.nola@suse.com>
Using the node external IP address for all CNI traffic is a breaking change from previous versions; we should make it an opt-in for distributed clusters instead of default behavior.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
We should be reading from the hijacked bufio.ReaderWriter instead of
directly from the net.Conn. There is a race condition where the
underlying http handler may consume bytes from the hijacked request
stream, if it comes in the same packet as the CONNECT header. These
bytes are left in the buffered reader, which we were not using. This was
causing us to occasionally drop a few bytes from the start of the
tunneled connection's client data stream.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
If CCM and ServiceLB are both disabled, don't run the cloud-controller-manager at all;
this should provide the same CLI flag behavior as previous releases, and not create
problems when users disable the CCM but still want ServiceLB.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Requires tweaking existing method signature to allow specifying whether or not IPv6 addresses should be return URL-safe.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
cadvisor still doesn't pull stats via CRI yet, so we have to continue to use the deprecated arg.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Use same kubelet-preferred-address-types setting as RKE2 to improve reliability of the egress selector when using a HTTP proxy. Also, use BindAddressOrLoopback to ensure that the correct supervisor address is used when --bind-address is set.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
* Move startup hooks wg into a runtime pointer, check before notifying systemd
* Switch default systemd notification to server
* Add 1 sec delay to allow etcd to write to disk
Signed-off-by: Derek Nola <derek.nola@suse.com>
The control-plane context handles requests outside the cluster and
should not be sent to the proxy.
In agent mode, we don't watch pods and just direct-dial any request for
a non-node address, which is the original behavior.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Watching pods appears to be the most reliable way to ensure that the
proxy routes and authorizes connections.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
This reverts commit aa9065749c.
Setting dual-stack node-ip does not work when --cloud-provider is set
to anything, including 'external'. Just set node-ip to the first IP, and
let the cloud provider add the other address.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
The cloud-provider arg is deprecated and cannot be set to anything other than external, but must still be used or node addresses are not set properly.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Reduces code complexity a bit and ensures we don't have to handle closed watch channels on our own
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Before this change, kube-router was always assuming that IPv4 is
enabled, which is not the case in IPv6-only clusters. To enable network
policies in IPv6-only, we need to explicitly let kube-router know when
to disable IPv4.
Signed-off-by: Michal Rostecki <vadorovsky@gmail.com>
* Bump etcd to v3.5.4-k3s1
* Fix issue with datastore corruption on cluster-reset
* Disable unnecessary components during cluster reset
Disable control-plane components and the tunnel setup during
cluster-reset, even when not doing a restore. This reduces the amount of
log clutter during cluster reset/restore, making any errors encountered
more obvious.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Problem:
Specifying extra arguments for the API server for example is not supported as
the arguments get stored in a map before being passed to the API server.
Solution:
Updated the GetArgs function to store the arguments in a map that can have
multiple values. Some more logic is added so that repeated extra arguments
retain their order when sorted whilst overall the arguments can still be
sorted for improved readability when logged.
Support has been added for prefixing and suffixing default argument values
by using -= and += when specifying extra arguments.
Signed-off-by: Terry Cain <terry@terrys-home.co.uk>
This is required to make the websocket tunnel server functional on
etcd-only nodes, and will save some code on the RKE2 side once pulled
through.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Closing idle connections isn't guaranteed to close out a pooled connection to a
loadbalancer endpoint that has been removed. Instead, ensure that requests used
to wait for the apiserver to become ready aren't reused.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Reuse the existing etcd library code to start up the temporary etcd
server for bootstrap reconcile. This allows us to do proper
health-checking of the datastore on startup, including handling of
alarms.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Several types contained redundant references to ControlRuntime data. Switch to consistently accessing this via config.Runtime instead.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
* Regular CLI framework for encrypt commands
* New secrets-encryption feature
* New integration test
* fixes for flaky integration test CI
* Fix to bootstrap on restart of existing nodes
* Consolidate event recorder
Signed-off-by: Derek Nola <derek.nola@suse.com>
* Add etcd extra args support for K3s
Signed-off-by: Chris Kim <oats87g@gmail.com>
* Add etcd custom argument integration test
Signed-off-by: Chris Kim <oats87g@gmail.com>
* go generate
Signed-off-by: Chris Kim <oats87g@gmail.com>
Since we now start the server's agent sooner and in the background, we
may need to wait longer than 30 seconds for the apiserver to become
ready on downstream projects such as RKE2.
Since this essentially just serves as an analogue for the server's
apiReady channel, there's little danger in setting it to something
relatively high.
Signed-off-by: Brad Davidson <brad.davidson@rancher.com>
Brad Davidson