* Refactor webhook framework to allow webhook define their flags
Pulled out of `launcher.go` the flag parsing logic and moved it into the `Main` methods of the webhooks (under `controller/cmd/proxy.injector/main.go` and `controller/cmd/sp-validator/main.go`), so that individual webhooks themselves can define the flags they want to use.
Also no longer require that webhooks have cluster-wide access.
Finally, renamed the type `webhook.handlerFunc` to `webhook.Handler` so it can be exported. This will be used in the upcoming jaeger webhook.
* Remove dependency of linkerd-config for most control plane components
This PR removes the dependency of `linkerd-config` into control
plane components by making all that information passed through CLI
flags. As most of these components require a couple of flags, passing
them as flags could be more helpful, as updations to the flags trigger a
rollout unlike a configMap update.
This does not update the proxy-injector as it needs a lot more data
and mounting `linkerd-config` is better.
Fixes#4191#4993
This bumps Kubernetes client-go to the latest v0.19.2 (We had to switch directly to 1.19 because of this issue). Bumping to v0.19.2 required upgrading to smi-sdk-go v0.4.1. This also depends on linkerd/stern#5
This consists of the following changes:
- Fix ./bin/update-codegen.sh by adding the template path to the gen commands, as it is needed after we moved to GOMOD.
- Bump all k8s related dependencies to v0.19.2
- Generate CRD types, client code using the latest k8s.io/code-generator
- Use context.Context as the first argument, in all code paths that touch the k8s client-go interface
Signed-off-by: Tarun Pothulapati <tarunpothulapati@outlook.com>
When the service-mirror component can't reach the target's k8s API, the goroutine blocks and it can't be unblocked.
This was happenining specifically in the case of the multicluster integration test (still to be pushed), where the source and target clusters are created in quick succession and the target's API service doesn't always have time to be exposed before being requested by the service mirror.
The fix consists on no longer have restartClusterWatcher be side-effecting, and instead return an error. If such error is not nil then the link watcher is stopped and reset after 10 seconds.
All of the code for the service mirror controller lives in the `linkerd/linkerd2/controller/cmd` package. It is typical for control plane components to only have a `main.go` entrypoint in the cmd package. This can sometimes make it hard to find the service mirror code since I wouldn't expect it to be in the cmd package.
We move the majority of the code to a dedicated controller package, leaving only main.go in the cmd package. This is purely organizational; no behavior change is expected.
Signed-off-by: Alex Leong <alex@buoyant.io>
## What/How
@adleong pointed out in #4780 that when enabling slices during an upgrade, the new value does not persist in the `linkerd-config` ConfigMap. I took a closer look and it seems that we were never overwriting the values in case they were different.
* To fix this, I added an if block when validating and building the upgrade options -- if the current flag value differs from what we have in the ConfigMap, then change the ConfigMap value.
* When doing so, I made sure to check that if the cluster does not support `EndpointSlices` yet the flag is set to true, we will error out. This is done similarly (copy&paste similarily) to what's in the install part.
* Additionally, I have noticed that the helm ConfigMap template stored the flag value under `enableEndpointSlices` field name. I assume this was not changed in the initial PR to reflect the changes made in the protocol buffer. The API (and thus the CLI) uses the field name `endpointSliceEnabled` instead. I have changed the config template so that helm installations will use the same field, which can then be used in the destination service or other components that may implement slice support in the future.
Signed-off-by: Matei David <matei.david.35@gmail.com>
This PR corrects misspellings identified by the [check-spelling action](https://github.com/marketplace/actions/check-spelling).
The misspellings have been reported at aaf440489e (commitcomment-41423663)
The action reports that the changes in this PR would make it happy: 5b82c6c5ca
Note: this PR does not include the action. If you're interested in running a spell check on every PR and push, that can be offered separately.
Signed-off-by: Josh Soref <jsoref@users.noreply.github.com>
Fixes#4774
When a service mirror controller is unable to connect to the target cluster's API, the service mirror controller crashes with the error that it has failed to sync caches. This error lacks the necessary detail to debug the situation. Unfortunately, client-go does not surface more useful information about why the caches failed to sync.
To make this more debuggable we do a couple things:
1. When creating the target cluster api client, we eagerly issue a server version check to test the connection. If the connection fails, the service-mirror-controller logs now look like this:
```
time="2020-07-30T23:53:31Z" level=info msg="Got updated link broken: {Name:broken Namespace:linkerd-multicluster TargetClusterName:broken TargetClusterDomain:cluster.local TargetClusterLinkerdNamespace:linkerd ClusterCredentialsSecret:cluster-credentials-broken GatewayAddress:35.230.81.215 GatewayPort:4143 GatewayIdentity:linkerd-gateway.linkerd-multicluster.serviceaccount.identity.linkerd.cluster.local ProbeSpec:ProbeSpec: {path: /health, port: 4181, period: 3s} Selector:{MatchLabels:map[] MatchExpressions:[{Key:mirror.linkerd.io/exported Operator:Exists Values:[]}]}}"
time="2020-07-30T23:54:01Z" level=error msg="Unable to create cluster watcher: cannot connect to api for target cluster remote: Get \"https://36.199.152.138/version?timeout=32s\": dial tcp 36.199.152.138:443: i/o timeout"
```
This error also no longer causes the service mirror controller to crash. Updating the Link resource will cause the service mirror controller to reload the credentials and try again.
2. We rearrange the checks in `linkerd check --multicluster` to perform the target API connectivity checks before the service mirror controller checks. This means that we can validate the target cluster API connection even if the service mirror controller is not healthy. We also add a server version check here to quickly determine if the connection is healthy. Sample check output:
```
linkerd-multicluster
--------------------
√ Link CRD exists
√ Link resources are valid
* broken
W0730 16:52:05.620806 36735 transport.go:243] Unable to cancel request for promhttp.RoundTripperFunc
× remote cluster access credentials are valid
* failed to connect to API for cluster: [broken]: Get "https://36.199.152.138/version?timeout=30s": net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)
see https://linkerd.io/checks/#l5d-smc-target-clusters-access for hints
W0730 16:52:35.645499 36735 transport.go:243] Unable to cancel request for promhttp.RoundTripperFunc
× clusters share trust anchors
Problematic clusters:
* broken: unable to fetch anchors: Get "https://36.199.152.138/api/v1/namespaces/linkerd/configmaps/linkerd-config?timeout=30s": net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)
see https://linkerd.io/checks/#l5d-multicluster-clusters-share-anchors for hints
√ service mirror controller has required permissions
* broken
√ service mirror controllers are running
* broken
× all gateway mirrors are healthy
wrong number of (0) gateway metrics entries for probe-gateway-broken.linkerd-multicluster
see https://linkerd.io/checks/#l5d-multicluster-gateways-endpoints for hints
√ all mirror services have endpoints
‼ all mirror services are part of a Link
mirror service voting-svc-gke.emojivoto is not part of any Link
see https://linkerd.io/checks/#l5d-multicluster-orphaned-services for hints
```
Some logs from the underlying go network libraries sneak into the output which is kinda gross but I don't think it interferes too much with being able to understand what's going on.
Signed-off-by: Alex Leong <alex@buoyant.io>
Fixes#4707
In order to remove a multicluster link, we add a `linkerd multicluster unlink` command which produces the yaml necessary to delete all of the resources associated with a `linkerd multicluster link`. These are:
* the link resource
* the service mirror controller deployment
* the service mirror controller's RBAC
* the probe gateway mirror for this link
* all mirror services for this link
This command follows the same pattern as the `linkerd uninstall` command in that its output is expected to be piped to `kubectl delete`. The typical usage of this command is:
```
linkerd --context=source multicluster unlink --cluster-name=foo | kubectl --context=source delete -f -
```
This change also fixes the shutdown lifecycle of the service mirror controller by properly having it listen for the shutdown signal and exit its main loop.
A few alternative designs were considered:
I investigated using owner references as suggested [here](https://github.com/linkerd/linkerd2/issues/4707#issuecomment-653494591) but it turns out that owner references must refer to resources in the same namespace (or to cluster scoped resources). This was not feasible here because a service mirror controller can create mirror services in many different namespaces.
I also considered having the service mirror controller delete the mirror services that it created during its own shutdown. However, this could lead to scenarios where the controller is killed before it finishes deleting the services that it created. It seemed more reliable to have all the deletions happen from `kubectl delete`. Since this is the case, we avoid having the service mirror controller delete mirror services, even when the link is deleted, to avoid the race condition where the controller and CLI both attempt to delete the same mirror services and one of them fails with a potentially alarming error message.
Signed-off-by: Alex Leong <alex@buoyant.io>
This PR removes the service mirror controller from `linkerd mc install` to `linkerd mc link`, as described in https://github.com/linkerd/rfc/pull/31. For fuller context, please see that RFC.
Basic multicluster functionality works here including:
* `linkerd mc install` installs the Link CRD but not any service mirror controllers
* `linkerd mc link` creates a Link resource and installs a service mirror controller which uses that Link
* The service mirror controller creates and manages mirror services, a gateway mirror, and their endpoints.
* The `linkerd mc gateways` command lists all linked target clusters, their liveliness, and probe latences.
* The `linkerd check` multicluster checks have been updated for the new architecture. Several checks have been rendered obsolete by the new architecture and have been removed.
The following are known issues requiring further work:
* the service mirror controller uses the existing `mirror.linkerd.io/gateway-name` and `mirror.linkerd.io/gateway-ns` annotations to select which services to mirror. it does not yet support configuring a label selector.
* an unlink command is needed for removing multicluster links: see https://github.com/linkerd/linkerd2/issues/4707
* an mc uninstall command is needed for uninstalling the multicluster addon: see https://github.com/linkerd/linkerd2/issues/4708
Signed-off-by: Alex Leong <alex@buoyant.io>
* Removes/Relaxes prometheus related checks
Now that prometheus is an add-on, There can be cases where prometheus is
disabled at which the check should show a warning but not fail. This
decouples the tight depedency.
This changes the following checks:
- Removes serviceAccount and pod checks in the CLI.
- Relaxes `linkerd-api` checks to only check for prometheus access when
the URL is not empty. This should work seamlessly with external
prometheus as that URL will be passed and it performs the same
check.
Signed-off-by: Tarun Pothulapati <tarunpothulapati@outlook.com>
EndpointSlices have been made opt-in due to their experimental nature. This PR
introduces a new install flag 'enableEndpointSlices' that will allow adopters to
specify in their cli install or helm install step whether they would like to
use endpointslices as a resource in the destination service, instead of the
endpoints k8s resource.
Signed-off-by: Matei David <matei.david.35@gmail.com>
Introduce support for the EndpointSlice k8s resource (k8s v1.16+) in the destination service.
Through this PR, in the EndpointsWatcher, there will be a dedicated informer for EndpointSlice;
the informer cannot run at the same time as the Endpoints resource informer. The main difference
is that EndpointSlices have a one-to-many relationship with a service, they provide better performance benefits,
dual-stack addresses and more. EndpointSlice support also implies service topology and other k8s related features.
Validated and tested manually, as well as with dedicated unit tests.
Closes#4501
Signed-off-by: Matei David <matei.david.35@gmail.com>
Fixes#4582
When a target cluster gateway is exposed as a hostname rather than with a fixed IP address, the service mirror controller fails to create mirror services and gateway mirrors for that gateway. This is because we only look at the IP field of the gateway service.
We make two changes to address this problem:
First, when extracting the gateway spec from a gateway that has a hostname instead of an IP address, we do a DNS lookup to resolve that hostname into an IP address to use in the mirror service endpoints and gateway mirror endpoints.
Second, we schedule a repair job on a regular (1 minute) to update these endpoint objects. This has the effect of re-resolving the DNS names every minute to pick up any changes in DNS resolution.
Signed-off-by: Alex Leong <alex@buoyant.io>
This PR just modifies the log levels on the probe and cluster watchers
to emit in INFO what they would emit in DEBUG. I think it makes sense
as we need that information to track problems. The only difference is
that when probing gateways we only log if the probe attempt was
unsuccessful.
Fix#4546
When the identity annotation on a gateway service is updated, this change is not propagated to the mirror gateway endpoints object.
This is because the annotations are updated on the wrong object and the changes are lost.
Signed-off-by: Alex Leong <alex@buoyant.io>
Change terminology from local/remote to source/target in events and metrics.
This does not change any variable, function, struct, or field names since
testing is still improving
Signed-off-by: Kevin Leimkuhler <kevin@kleimkuhler.com>
This change modifies the linkerd-gateway component to use the inbound
proxy, rather than nginx, for gateway. This allows us to detect loops and
propagate identity through the gateway.
This change also cleans up port naming to `mc-gateway` and `mc-probe`
to resolve conflicts with Kubernetes validation.
---
* proxy: v2.99.0
The proxy can now operate as gateway, routing requests from its inbound
proxy to the outbound proxy, without passing the requests to a local
application. This supports Linkerd's multicluster feature by adding a
`Forwarded` header to propagate the original client identity and assist
in loop detection.
---
* Add loop detection to inbound & TCP forwarding (linkerd/linkerd2-proxy#527)
* Test loop detection (linkerd/linkerd2-proxy#532)
* fallback: Unwrap errors recursively (linkerd/linkerd2-proxy#534)
* app: Split inbound/outbound constructors into components (linkerd/linkerd2-proxy#533)
* Introduce a gateway between inbound and outbound (linkerd/linkerd2-proxy#540)
* gateway: Add a Forwarded header (linkerd/linkerd2-proxy#544)
* gateway: Return errors instead of responses (linkerd/linkerd2-proxy#547)
* Fail requests that loop through the gateway (linkerd/linkerd2-proxy#545)
* inject: Support config.linkerd.io/enable-gateway
This change introduces a new annotation,
config.linkerd.io/enable-gateway, that, when set, enables the proxy to
act as a gateway, routing all traffic targetting the inbound listener
through the outbound proxy.
This also removes the nginx default listener and gateway port of 4180,
instead using 4143 (the inbound port).
* proxy: v2.100.0
This change modifies the inbound gateway caching so that requests may be
routed to multiple leaves of a traffic split.
---
* inbound: Do not cache gateway services (linkerd/linkerd2-proxy#549)
Change terminology from local/remote to source/target in service-mirror and
healthchecks help text.
This does not change any variable, function, struct, or field names since
testing is still improving
Signed-off-by: Kevin Leimkuhler <kevin@kleimkuhler.com>
There are a few notable things happening in this PR:
- the probe manager has been decoupled from the cluster_watcher. Now its only responsibility is to watch for mirrored gateways beeing created and to probe them. This means that probes are initiated for all gateways no matter whether there are mirrored services being paired
- the number of paired services is derived from the existing services in the cluster rather than being published as a metric by the prober
- there are no events being exchanged between the cluster watcher and the probe manager
Signed-off-by: Zahari Dichev <zaharidichev@gmail.com>
THis PR addresses two problems:
- when a resync happens (or the mirror controller is restarted) we incorrectly classify the remote gateway as a mirrored service that is not mirrored anymore and we delete it
- when updating services due to a gateway update, we need to select only the services for the particular cluster
The latter fixes#4451
This change creates a gateway proxy for every gateway. This enables the probe worker to leverage the destination service functionality in order to discover the identity of the gateway.
Fix#4411
Signed-off-by: Zahari Dichev <zaharidichev@gmail.com>
This PR introduces a few changes that were requested after a bit of service mirror reviewing.
- we restrict the RBACs so the service mirror controller cannot read secrets in all namespaces but only in the one that it is installed in
- we unify the namespace namings so all multicluster resources are installedi n `linkerd-multicluster` on both clusters
- fixed checks to account for changes
Signed-off-by: Zahari Dichev <zaharidichev@gmail.com>
Unit tests that exercise most of the code in cluster_watcher.go. Essentially the whole cluster mirroring machinary can be tought of as a function that takes remote cluster state, local cluster state, and modification events and as a result it either modifies local cluster state or issues new events onto the queue. This is what these tests are trying to model. I think this covers a lot of the logic there. Any suggestions for other edge cases are welcome.
Signed-off-by: Zahari Dichev <zaharidichev@gmail.com>
This PR introduces a service mirroring component that is responsible for watching remote clusters and mirroring their services locally.
Signed-off-by: Zahari Dichev <zaharidichev@gmail.com>
Fixes#3444Fixes#3443
## Background and Behavior
This change adds support for the destination service to resolve Get requests which contain a service clusterIP or pod ip as the `Path` parameter. It returns the stream of endpoints, just as if `Get` had been called with the service's authority. This lays the groundwork for allowing the proxy to TLS TCP connections by allowing the proxy to do destination lookups for the SO_ORIG_DST of tcp connections. When that ip address corresponds to a service cluster ip or pod ip, the destination service will return the endpoints stream, including the pod metadata required to establish identity.
Prior to this change, attempting to look up an ip address in the destination service would result in a `InvalidArgument` error.
Updating the `GetProfile` method to support ip address lookups is out of scope and attempts to look up an ip address with the `GetProfile` method will result in `InvalidArgument`.
## Implementation
We do this by creating a `IPWatcher` which wraps the `EndpointsWatcher` and supports lookups by ip. `IPWatcher` maintains a mapping up clusterIPs to service ids and translates subscriptions to an IP address into a subscription to the service id using the underlying `EndpointsWatcher`.
Since the service name is no longer always infer-able directly from the input parameters, we restructure `EndpointTranslator` and `PodSet` so that we propagate the service name from the endpoints API response.
## Testing
This can be tested by running the destination service locally, using the current kube context to connect to a Kubernetes cluster:
```
go run controller/cmd/main.go destination -kubeconfig ~/.kube/config
```
Then lookups can be issued using the destination client:
```
go run controller/script/destination-client/main.go -path 192.168.54.78:80 -method get -addr localhost:8086
```
Service cluster ips and pod ips can be used as the `path` argument.
Signed-off-by: Alex Leong <alex@buoyant.io>
This PR adds support for CronJobs and ReplicaSets to `linkerd inject`, the web
dashboard and CLI. It adds a new Grafana dashboard for each kind of resource.
Closes#3614Closes#3630Closes#3584Closes#3585
Signed-off-by: Sergio Castaño Arteaga tegioz@icloud.com
Signed-off-by: Cintia Sanchez Garcia cynthiasg@icloud.com
* Fixed bad identity string for target pod in tap
Fixes#3506
Was using the cluster domain instead of the trust domain, which results
in an error when those domains differ.
* If tap source IP matches many running pods then only show the IP
When an unmeshed source ip matched more than one running pod, tap was
showing the names for all those pods, even though the didn't necessary
originate the connection. This could be reproduced when using pod
network add-on such as Calico.
With this change, if a node matches, return it, otherwise we proceed to look for a matching pod. If exactly one running pod matches we return it. Otherwise we return just the IP.
Fixes#3103
* Add support for --identity-issuer-mode flag to install cmd
* Change flag to be a bool
* Read correct data form identity when external issuer is used
* Add ability for identity service to dynamically reload certs
* Fix failing tests
* Minor refactor
* Load trust anchors from identity issuer secret
* Make identity service actually watch for issuer certs updates
* Add some testing around cmd line identity options validation
* Add tests ensuring that identity service loads issuer
* Take into account external-issuer flag during upgrade + tests
* Fix failing upgrade test
* Address initial review feedback
* Address further review feedback on cli and helm
* Do not persist --identity-external-issuer
* Some improvements to identitiy service
* Bring back persistane of external issuer flag
* Address more feedback
* Update dockerfiles shas
* Publishing k8s events on issuer certs rotation
* Ensure --ignore-cluster+external issuer is not supported
* Update go-deps shas
* Transition to identity issuer scheme based configuration
* Use k8s consts for secret file names
Signed-off-by: zaharidichev <zaharidichev@gmail.com>
When running the destination controller locally, the Linkerd config files which are typically mounted from a configmap are not available. To facilitate local development, we fall back to default values in this case instead of failing to start up.
Signed-off-by: Alex Leong <alex@buoyant.io>
Followup to #2990, which refactored `linkerd endpoints` to use the
`Destination.Get` API instead of the `Discovery.Endpoints` API, leaving
the Discovery with no implented methods. This PR removes all the Discovery
code leftovers.
Fixes#3499
This reverts commit edd3b1f6d4.
This is a temporary revert of #3461 while we sort out some details of how this should configured and how it should interact with configuring a trace collector on the Linkerd proxy. We will reintroduce this change once the config plan is straightened out.
Signed-off-by: Alex Leong <alex@buoyant.io>
Alejandro Pedraza
Tarun Pothulapati
Alex Leong
Matei David
Josh Soref
Arthur Silva Sens
Zahari Dichev
Kevin Leimkuhler
Oliver Gould
Mayank Shah
Sergio C. Arteaga