This branch removes the `--proxy-bind-timeout` flag from the
`linkerd inject` and `linkerd install` CLI commands, and the
`LINKERD2_PROXY_BIND_TIMEOUT` environment variable from their output.
This is in preparation for removing that timeout from the proxy (as
described in #2013).
I thought it was prudent to remove this from the CLIs before removing it
from the proxy, so we can't create a situation where the CLIs produce
output that results in broken proxy containers.
Fixes#2013
Signed-off-by: Eliza Weisman <eliza@buoyant.io>
When `GetProfiles` is called for a destination that does not have a service profile, the proxy-api service does not return any messages until a service profile is created for that service. This can be interpreted as hanging, and can make it difficult to calculate response latency metrics.
Change the behavior of the API to always return a service profile message immediately. If the service does not have a service profile, the default service profile is returned.
Signed-off-by: Alex Leong <alex@buoyant.io>
DaemonSet stats are not currently shown in the cli stat command, web ui
or grafana dashboard. This commit adds daemonset support for stat.
Update stat command's help message to reference daemonsets.
Update the public-api to support stats for daemonsets.
Add tests for stat summary and api.
Add daemonset get/list/watch permissions to the linkerd-controller
cluster role that's created using the install command.
Update golden expectation test files for install command
yaml manifest output.
Update web UI with daemonsets
Update navigation, overview and pages to list daemonsets and the pods
associated to them.
Add daemonset paths to server, and ui apps.
Add grafana dashboard for daemonsets; a clone of the deployment
dashboard.
Update dependencies and dockerfile hashes
Add DaemonSet support to tap and top commands
Fixes of #2006
Signed-off-by: Zak Knill <zrjknill@gmail.com>
Fixes#2119
When Linkerd is installed in single-namespace mode, the public-api container panics when it attempts to access watch service profiles.
In single-namespace mode, we no longer watch service profiles and return an informative error when the TopRoutes API is called.
Signed-off-by: Alex Leong <alex@buoyant.io>
* Introduce resource selector and deprecate namespace field for ListPods
* Changes from code review
* Properly deprecate the field
* Do not check for nil
* Fix the mockProm usage
* Protoc changes revert
* Changed from code review
Signed-off-by: Alena Varkockova <varkockova.a@gmail.com>
When debugging control plane issues or issues pertaining to a linkerd proxy, it can be cumbersome to get logs from affected containers quickly.
This PR adds a new `logs` command to the Linkerd CLI to surface log lines from any container within linkerd's control plane. This feature relies heavily on [stern](https://github.com/wercker/stern), which already provides this behavior. This PR integrates this package into the Linkerd CLI to allow users to quickly retrieve logs whenever they run into issues when using Linkerd.
Signed-off-by: Dennis Adjei-Baah <dennis@buoyant.io>
Fixes#1875
This change improves the `linkerd routes` command in a number of important ways:
* The restriction on the type of the `--to` argument is lifted and any resource type can now be used. Try `--to ns/books`, `--to po/webapp-ABCDEF`, `--to au/linkerd.io`, or even `--to svc`.
* All routes for the target will now be populated in the table, even if there are no Prometheus metrics for that route.
* [UNKNOWN] has been renamed to [DEFAULT]
* The `Service/Authority` column will now list `Service` in all cases except for when an authority target is explicitly requested.
```
$ linkerd routes deploy/traffic --to deploy/webapp
ROUTE SERVICE SUCCESS RPS LATENCY_P50 LATENCY_P95 LATENCY_P99
GET / webapp 100.00% 0.5rps 50ms 180ms 196ms
GET /authors/{id} webapp 100.00% 0.5rps 100ms 900ms 980ms
GET /books/{id} webapp 100.00% 0.9rps 38ms 93ms 99ms
POST /authors webapp 100.00% 0.5rps 35ms 48ms 50ms
POST /authors/{id}/delete webapp 100.00% 0.5rps 83ms 180ms 196ms
POST /authors/{id}/edit webapp 0.00% 0.0rps 0ms 0ms 0ms
POST /books webapp 45.16% 2.1rps 75ms 425ms 485ms
POST /books/{id}/delete webapp 100.00% 0.5rps 30ms 90ms 98ms
POST /books/{id}/edit webapp 56.00% 0.8rps 92ms 875ms 975ms
[DEFAULT] webapp 0.00% 0.0rps 0ms 0ms 0ms
```
This is all made possible by a shift in the way we handle the destination resource. When we get a request with a `ToResource`, we use the k8s API to find all Services which include at least one pod belonging to that resource. We then fetch all service profiles for those services and display the routes from those serivce profiles.
This shift in thinking also precipitates a change in the TopRoutes API where we no longer need special cases for `ToAll` (which can be specified by `--to au`) or `ToAuthority` (which can be specified by `--to au/<authority>`) and instead can use a `ToResource` to handle all cases.
Signed-off-by: Alex Leong <alex@buoyant.io>
The outputs of the `check` and `inject` commands did not vary much
between successful and failed executions, and were a bit verbose and
challenging to parse.
Reorganize output of `check` and `inject` commands, to provide more
output when errors occur, and less output when successful.
Specific changes:
`linkerd check`
- visually group checks by category
- introduce `hintURL`'s, to provide doc links when checks fail
- add spinners when retrying, remove additional retry lines
- colored unicode characters to indicate success/warning/failure
`linkerd inject`
- modify default output to mirror `kubectl apply`
- only output non-successful inject reports
- support `--verbose` flag to output all inject reports
Fixes#1471, #1653, #1656, #1739
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Setup port-forwarding for linkerd dashboard command
* Output port-forward logs when --verbose flag is set
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
This is a followup branch from #2023:
- delete `proxy/client.go`, move code to `destination-client`
- move `RenderTapEvent` and stat functions from `util` to `cmd`
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
Commit 1: Enable lint check for comments
Part of #217. Follow up from #1982 and #2018.
A subsequent commit will fix the ci failure.
Commit 2: Address all comment-related linter errors.
This change addresses all comment-related linter errors by doing the
following:
- Add comments to exported symbols
- Make some exported symbols private
- Recommend via TODOs that some exported symbols should should move or
be removed
This PR does not:
- Modify, move, or remove any code
- Modify existing comments
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* add securityContext with runAsUser: {{.ProxyUID}} to the various containers in the install template
* Update golden to reflect new additions
* changed to a different user id than the proxy user id
* Added a controller-uid install option
* change the port that the proxy-injector runs
* The initContainers needs to be run as the root user.
* move security contexts to container level
Signed-off-by: Cody Vandermyn <cody.vandermyn@nordstrom.com>
* Add parameter to stats API to skip retrieving Prometheus stats
Used by the dashboard to populate list of resources.
Fixes#1022
Signed-off-by: Alejandro Pedraza <alejandro.pedraza@gmail.com>
* Prometheus queries check results were being ignored
* Refactor verifyPromQueries() to also test when no prometheus queries
should be generated
* Add test for SkipStats=true
Includes adding ability to public.GenStatSummaryResponse to not generate
basicStats
* Fix previous test
Rename snake case fields to camel case in service profile spec. This improves the way they are rendered when the `kubectl describe` command is used.
Signed-off-by: Alex Leong <alex@buoyant.io>
Add support for service profiles created on external (non-service) authorities. For example, this allows you to create a service profile named `linkerd.io` which will apply to calls made to `linkerd.io`.
This is done by changing the `LINKERD2_PROXY_DESTINATION_PROFILE_SUFFIXES` to `.` so that the proxy will attempt to lookup a service profile for any authority. We provide the `--disable-external-profiles` proxy flag to revert this behavior in case it is a problem.
We also refactor the proxy-api implementation of GetProfiles so that it does the profile lookup, regardless of if the authority looks like a Kubernetes service name or not. To simplify this, support for multiple resolves (which was unused) was removed.
Signed-off-by: Alex Leong <alex@buoyant.io>
When debugging issues, it's helpful to disable HTTP/2 upgrading to
simplify diagnostics.
This chagne adds an `enable-h2-ugprade` flag to _proxy-api_. When this
flag is set to false, the proxy-api will not suggest that meshed
endpoints are upgraded to use HTTP/2.
As a follow-up, a flag should be added to `install` to control how the
proxy-api is initialized.
We rework the routes command so that it can accept any Kubernetes resource, making it act much more similarly to the stat command.
Signed-off-by: Alex Leong <alex@buoyant.io>
We rename path to path_regex in the ServiceProfile CRD to make it clear that this field accepts a regular expression. We also take this opportunity to remove unnecessary line anchors from regular expressions now that these anchors are added in the proxy.
Signed-off-by: Alex Leong <alex@buoyant.io>
Filtering by Kubernetes job was not supported. Also filtering by any unknown
type caused a panic.
Add filtering support by Kubernetes job, with special case mapping `job` to
`k8s_job`, to not conflict with Prometheus' job label.
Fix panic when unknown type specified as a `--from` or `--to` flag.
Fix `job` label from `linkerd-proxy` overwriting Prometheus `job` label at
collection time. This caused all metrics collected by proxy sidecars in
Kubernetes jobs to be collected into an incorrect Prometheus job, rather than
the expected `linkerd-proxy` Prometheus job.
Fix `unsupported resource type` tap error message incorrectly printing the
target resource rather than the destination.
Set `--controller-log-level debug` in `install_test.go` for easier debugging.
Expose `slow-cooker`'s metrics via a k8s service in the tap integration test, to
validate proxy requests with a job as destination.
Fixes#1872
Part of #627
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
This change alters the controller's Tap service to include route labels
when translating tap events, modifies the public API to include route
metadata in responses, and modifies the tap CLI command to include
rt_ labels in tap output (when -o wide is used).
* Adjust proxy, Prometheus, and Grafana probes
High `readinessProbe.initialDelaySeconds` values delayed the controller's
readiness by up to 30s, preventing cli commands from succeeding shortly after
control plane deployment.
Decrease `readinessProbe.initialDelaySeconds` in the proxy, Prometheus, and
Grafana to the default 0s. Also change `linkerd check` controller pod ordering
to: controller, prometheus, web, grafana.
Detailed probe changes:
- proxy
- decrease `readinessProbe.initialDelaySeconds` from 10s to 0s
- prometheus
- decrease `readinessProbe.initialDelaySeconds` from 30s to 0s
- decrease `readinessProbe.timeoutSeconds` from 30s to 1s
- decrease `livenessProbe.timeoutSeconds` from 30s to 1s
- grafana
- decrease `readinessProbe.initialDelaySeconds` from 30s to 0s
- decrease `readinessProbe.timeoutSeconds` from 30s to 1s
- decrease `readinessProbe.failureThreshold` from 10 to 3
- increase `livenessProbe.initialDelaySeconds` from 0s to 30s
Fixes#1804
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
The `--open-api` flag is an alternative to the `--template` flag for the `linkerd profile` command. It reads an OpenAPI specification file (also called a swagger file) and uses it to generate a corresponding service profile.
Signed-off-by: Alex Leong <alex@buoyant.io>
When using `--proxy-auto-inject` with Kuberntes `v1.9.11`, observed auto
injector incorrectly merging list elements rather than inserting new
ones. This issue was not reproducible on `v1.10.3`.
For example, this input:
```
spec:
template:
spec:
containers:
- name: vote-bot
command:
- emojivoto-vote-bot
```
Would yield:
```
spec:
template:
spec:
containers:
- name: linkerd-proxy
command:
- emojivoto-vote-bot
- name: vote-bot
command:
- emojivoto-vote-bot
```
This change replaces json patch specs like
`/spec/template/spec/containers/0` with
`/spec/template/spec/containers/-`. The former is intended to insert at
the beggining of a list, the latter at the end. This also simplifies the
code a bit and more closely aligns with the intent of injecting at the
end of lists.
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
Add a barebones ListServices endpoint, in support of autocomplete for services.
As we develop service profiles, this endpoint could probably be used to describe
more aspects of services (like, if there were some way to check whether a
service profile was enabled or not).
Accessible from the web UI via http://localhost:8084/api/services
The `linkerd` routes command only supports outbound metrics queries (i.e. ones with the `--from` flag). Inbound queries (i.e. ones without the `--from` flag) never return any metrics.
We update the proxy version and use the new canonicalized form for dst labels to gain support for inbound metrics as well.
Signed-off-by: Alex Leong <alex@buoyant.io>
The tap server accesses protobuf fields directly instead of using the
`Get*()` accessors. The accessors are necessary to prevent dereferencing
a nil pointer and crashing the tap service.
Furthermore, these maps are explicitly initialized when `nil` to support
label hydration.
Add a routes command which displays per-route stats for services that have service profiles defined.
This change has three parts:
* A new public-api RPC called `TopRoutes` which serves per-route stat data about a service
* An implementation of TopRoutes in the public-api service. This implementation reads per-route data from Prometheus. This is very similar to how the StatSummaries RPC and much of the code was able to be refactored and shared.
* A new CLI command called `routes` which displays the per-route data in a tabular or json format. This is very similar to the `stat` command and much of the code was able to be refactored and shared.
Note that as of the currently targeted proxy version, only outbound route stats are supported so the `--from` flag must be included in order to see data. This restriction will be lifted in an upcoming change once we add support for inbound route stats as well.
Signed-off-by: Alex Leong <alex@buoyant.io>
# Problem
When we add a `--from` query to `linkerd stat au` we get more rows than if we would have just run `linkerd stat au`.
Adding a `--from` causes an extra row to be added, and the named authority to be ignored (this is the result we would have expected when running `linkerd stat au -n emojivoto --from deploy/web`).
# Solution
Destination query labels are now appended to `labels` so that those labels can be filtered on.
# Validation
Tests have been updated to reflect the expected expected destination labels now appended in `--from` queries.
Fixes#1766
Signed-off-by: Kevin Leimkuhler <kevin@kleimkuhler.com>
* Refactor util.BuildResource so it can deal with multiple resources
First step to address #1487: Allow stat summary to query for multiple
resources
Signed-off-by: Alejandro Pedraza <alejandro.pedraza@gmail.com>
* Update the stat cli help text to explain the new multi resource querying ability
Propsal for #1487: Allow stat summary to query for multiple resources
Signed-off-by: Alejandro Pedraza <alejandro.pedraza@gmail.com>
* Allow stat summary to query for multiple resources
Implement this ability by issuing parallel requests to requestStatsFromAPI()
Proposal for #1487
Signed-off-by: Alejandro Pedraza <alejandro.pedraza@gmail.com>
* Update tests as part of multi-resource support in `linkerd stat` (#1487)
- Refactor stat_test.go to reuse the same logic in multiple tests, and
add cases and files for json output.
- Add a couple of cases to api_utils_test.go to test multiple resources
validation.
Signed-off-by: Alejandro Pedraza <alejandro.pedraza@gmail.com>
* `linkerd stat` called with multiple resources should keep an ordering (#1487)
Add SortedRes holding the order of resources to be followed when
querying `linkerd stat` with multiple resources
Signed-off-by: Alejandro Pedraza <alejandro.pedraza@gmail.com>
* Extra validations for `linkerd stat` with multiple resources (#1487)
Signed-off-by: Alejandro Pedraza <alejandro.pedraza@gmail.com>
* `linkerd stat` resource grouping, ordering and name prefixing (#1487)
- Group together stats per resource type.
- When more than one resource, prepend name with type.
- Make sure tables always appear in the same order.
Signed-off-by: Alejandro Pedraza <alejandro.pedraza@gmail.com>
* Allow `linkerd stat` to be called with multiple resources
A few final refactorings as per code review.
Fixes#1487
Signed-off-by: Alejandro Pedraza <alejandro.pedraza@gmail.com>
This commit removes duplicate logic that loads Kubernetes config and
replaces it with GetConfig from pkg/k8s. This also allows to load
config from default sources like $KUBECONFIG instead of explicitly
passing -kubeconfig option to controller components.
Signed-off-by: Igor Zibarev <zibarev.i@gmail.com>
We make several changes to the service profile spec to make service profiles more ergonomic and to make them more consistent with the destination profile API.
* Allow multiple fields to be simultaneously set on a RequestMatch or ResponseMatch condition. Doing so is equivalent to combining the fields with an "all" condition.
* Rename "responses" to "response_classes"
* Change "IsSuccess" to "is_failure"
Signed-off-by: Alex Leong <alex@buoyant.io>
A container called `proxy-api` runs in the Linkerd2 controller pod. This container listens on port 8086 and serves the proxy-api but does nothing other than forward gRPC requests to the destination container which listens on port 8089.
We remove the proxy-api container altogether and change the destination container to listen on port 8086 instead of 8089. The result is that clients still use the proxy-api by connecting to `proxy-api.<ns>.svc.cluster.local:8086` but the controller has one fewer containers. This results in a simpler system that is easier to reason about.
Signed-off-by: Alex Leong <alex@buoyant.io>
Service profiles must be named in the form `"<service>.<namespace>"`. This is inconsistent with the fully normalized domain name that the proxy sends to the controller. It also does not permit creating service profiles for non-Kubernetes services.
We switch to requiring that service profiles must be named with the FQDN of their service. For Kubernetes services, this is `"<service>.<namespace>.svc.cluster.local"`.
This change alone is not sufficient for allowing service profile for non-Kubernetes services because the k8s resolver will ignore any DNS names which are not Kubernetes services. Further refactoring of the resolver will be required to allow looking up non-Kubernetes service profiles in Kuberenetes.
Signed-off-by: Alex Leong <alex@buoyant.io>
The `proxy-api` service included a stub implementation of `GetProfile`
instead of forwarding requests to the `destination` service.
This change fills in the proxy-api service's `GetProfile` implementation
to forward requests to the destination service.
Fix broken docker build by moving Service Profile conversion and validation into `/pkg`.
Fix broken integration test by adding service profile validation output to `check`'s expected output.
Testing done:
* `gotest -v ./...`
* `bin/docker-build`
* `bin/test-run (pwd)/bin/linkerd`
Signed-off-by: Alex Leong <alex@buoyant.io>
Add a check to `linkerd check` which validates all service profile resources. In particular it checks:
* does the service profile refer to an existent service
* is the service profile valid
Signed-off-by: Alex Leong <alex@buoyant.io>
We implement the getProfiles method in the destination service. This method returns a stream of destination profiles for a given authority. It does this by looking up the ServiceProfile resource in the controller namespace named `<svc>.<ns>` where `<svc>` is the name of the service and `<ns>` is the namespace of the service.
This PR includes:
* Adding a ServiceProfile Custom Resource Definition to linkerd install
* A watch based implementation of the getProfiles method in the destination service, similar to the implementation of get.
* An update to the destination client script that allows querying the getProfiles method.
Signed-off-by: Alex Leong <alex@buoyant.io>
* Ensure that the proxy injector mutating webhook preserves the original labels
and annotations
The deployment's selector must also match the pod template labels in
newer version of Kubernetes.
This resolves issue #1756.
* Add the Linkerd labels to the deployment metadata during auto proxy
injection
* Remove selector match labels JSON patch from proxy injector
This isn't needed to resolve the selector label mismatch errors.
Signed-off-by: ihcsim <ihcsim@gmail.com>
Appending proxy-init to the end of the list ensures that it won't
interfere with other init containers from accessing the network,
before the proxy container is created.
This resolves bug #1760
Signed-off-by: ihcsim <ihcsim@gmail.com>
Added support for json output in `linkerd stat` through a new (-o|--output)=json option.
Fixes#1417
Signed-off-by: Alejandro Pedraza <alejandro.pedraza@gmail.com>
Updates to the Kubernetes utility code in `/controller/k8s` to support interacting with ServiceProfiles.
This makes use of the code generated client added in #1752
Signed-off-by: Alex Leong <alex@buoyant.io>
To support reading and writing of the ServiceProfile custom resource, we add a codegen'd Kubernetes client for this resource.
* Adding the ServiceProfile type and related boilerplate to /controller/gen/apis/serviceprofile. This boilerplate also contains directives that control how codegen works.
* A script in /hack which invokes codegen that generates Kubernetes client machinery for interacting with ServiceProfile resources. The majority of the generated code lives in /controller/gen/client.
* The above-mentioned generated code.
Signed-off-by: Alex Leong <alex@buoyant.io>
* Add --single-namespace install flag for restricted permissions
* Better formatting in install template
* Mark --single-namespace and --proxy-auto-inject as experimental
* Fix wording of --single-namespace check flag
* Small healthcheck refactor
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
`go test` was failing with
`Fatalf call has arguments but no formatting directives`
Fix test failure, make all logrus api calls consistent.
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Support auto sidecar-injection
1. Add proxy-injector deployment spec to cli/install/template.go
2. Inject the Linkerd CA bundle into the MutatingWebhookConfiguration
during the webhook's start-up process.
3. Add a new handler to the CA controller to create a new secret for the
webhook when a new MutatingWebhookConfiguration is created.
4. Declare a config map to store the proxy and proxy-init container
specs used during the auto-inject process.
5. Ignore namespace and pods that are labeled with
linkerd.io/auto-inject: disabled or linkerd.io/auto-inject: completed
6. Add new flag to `linkerd install` to enable/disable proxy
auto-injection
Proposed implementation for #561.
* Resolve missing packages errors
* Move the auto-inject label to the pod level
* PR review items
* Move proxy-injector to its own deployment
* Ignore pods that already have proxy injected
This ensures the webhook doesn't error out due to proxy that are injected using the command
* PR review items on creating/updating the MWC on-start
* Replace API calls to ConfigMap with file reads
* Fixed post-rebase broken tests
* Don't mutate the auto-inject label
Since we started using healhcheck.HasExistingSidecars() to ensure pods with
existing proxies aren't mutated, we don't need to use the auto-inject label as
an indicator.
This resolves a bug which happens with the kubectl run command where the deployment
is also assigned the auto-inject label. The mutation causes the pod auto-inject
label to not match the deployment label, causing kubectl run to fail.
* Tidy up unit tests
* Include proxy resource requests in sidecar config map
* Fixes to broken YAML in CLI install config
The ignore inbound and outbound ports are changed to string type to
avoid broken YAML caused by the string conversion in the uint slice.
Also, parameterized the proxy bind timeout option in template.go.
Renamed the sidecar config map to
'linkerd-proxy-injector-webhook-config'.
Signed-off-by: ihcsim <ihcsim@gmail.com>
Horizontal Pod Autoscaling does not work when container definitions in pods do not all have resource requests, so here's the ability to add CPU + Memory requests to install + inject commands by proving proxy options --proxy-cpu + --proxy-memory
Fixes#1480
Signed-off-by: Ben Lambert <ben@blam.sh>
* Use ListPods always for data plane HC
* Missing changes in grpc_server.go
* Address review comments
* Read proxy version from spec
Signed-off-by: Alena Varkockova <varkockova.a@gmail.com>
Add checks to `linkerd check --pre` to verify that the user has permission to create:
* namespaces
* serviceaccounts
* clusterroles
* clusterrolebindings
* services
* deployments
* configmaps
Signed-off-by: Alex Leong <alex@buoyant.io>
Sometimes, the tap server causes the controller pod to restart after it receives this error.
This error arises when the Tap server does not close gRPC tap streams to proxies before the tap server terminates its streams to its upstream clients and causes the controller pod to restart.
This PR uses the request context from the initial TapByReource to help shutdown tap streams to the data plane proxies gracefully.
fixes#1504
Signed-off-by: Dennis Adjei-Baah <dennis@buoyant.io>
If an input file is un-injectable, existing inject behavior is to simply
output a copy of the input.
Introduce a report, printed to stderr, that communicates the end state
of the inject command. Currently this includes checking for hostNetwork
and unsupported resources.
Malformed YAML documents will continue to cause no YAML output, and return
error code 1.
This change also modifies integration tests to handle stdout and stderr separately.
example outputs...
some pods injected, none with host networking:
```
hostNetwork: pods do not use host networking...............................[ok]
supported: at least one resource injected..................................[ok]
Summary: 4 of 8 YAML document(s) injected
deploy/emoji
deploy/voting
deploy/web
deploy/vote-bot
```
some pods injected, one host networking:
```
hostNetwork: pods do not use host networking...............................[warn] -- deploy/vote-bot uses "hostNetwork: true"
supported: at least one resource injected..................................[ok]
Summary: 3 of 8 YAML document(s) injected
deploy/emoji
deploy/voting
deploy/web
```
no pods injected:
```
hostNetwork: pods do not use host networking...............................[warn] -- deploy/emoji, deploy/voting, deploy/web, deploy/vote-bot use "hostNetwork: true"
supported: at least one resource injected..................................[warn] -- no supported objects found
Summary: 0 of 8 YAML document(s) injected
```
TODO: check for UDP and other init containers
Part of #1516
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
Increase the MaxRps on the tap server to 100 RPS.
The max RPS for tap/top was increased in for the CLI #1531, but we were
still manually setting this to 1 RPS in the Web UI and Web server.
Remove the pervasive setting of MaxRps to 1 in the web frontend and server
The default value for the max-rps argument to the tap and top commands is an overly conservative 1rps. This causes the data to come in very slowly and much data to be discarded. Furthermore, because tap requests are windowed to 10 seconds, this causes long pauses between updates.
We fix this in two ways. Firstly we reduce the window size to 1s so that updates will come in at least once per second, even when the actual RPS of the data path is extremely high. Secondly, we increase the default max-rps parameter from 1 to 100. This allows tap to paint an accurate picture of the data much more quickly and sidesteps some sampling bias that happens when the max-rps is low.
In general, tap events tend to happen in bursts. For example, one request in may trigger one or more requests out. Likewise, a single upstream event may trigger several requests to the tapped pod in quick succession. Sampling bias will occur when the max-rps is less than the actual rps and when the tap event limit subdivides these event bursts (biasing towards the first few events in the burst). The greater the max-rps, the less the effects of this bias.
Fixes#1525
Signed-off-by: Alex Leong <alex@buoyant.io>
Previously, we would tap any resource's pods, regardless of whether the pods
were meshed or not. We can't actually tap non-meshed pods, so I'm adding a check
that will filter out non-meshed pods from the pods that tap watches.
Previous behaviour:
When attempting to hang a non meshed pod, it would establish
a watch on the pods, but then never return any results. In the CLI you could
just cancel it with Ctrl-C. In the web, clicking Stop would send a
WebSocket.close(1000) but wouldn't actually close the connection...
Behaviour after change :
If no pods under the specified resource are meshed, it'll
return an error of no pods being found to tap
Fixes#1493.
When the tap server hydrates metadata for the source or destination peer
of a Tap event from the peer's IP address, it doesn't currently add a
namespace label. However, destinations labeled by the proxy do have such
a label.
This is because the tap server currently gets the hydrated labels from
the `GetPodLabels` function, which is also used by the Destination
service for labeling the individual endpoints in a `WeightedAddrSet`
response. However, the Destination service also adds some labels to all
the endpoints in the set, including the namespace and service, so
`GetPodLabels` doesn't return these labels. However, when the tap server
uses that function, it does not add the service or namespace labels.
This branch fixes this issue by adding those labels to the Tap event
after calling `GetPodLabels`. In addition, it fixes a missing space
between the `src/dst_res` and `src/dst_ns` labels in Tap CLI output
with the `-o wide` flag set. This issue was introduced during the
review of #1437, but was missed at the time because the namespace label
wasn't being set correctly.
Signed-off-by: Eliza Weisman <eliza@buoyant.io>
Closes#1170.
This branch adds a `-o wide` (or `--output wide`) flag to the Tap CLI.
Passing this flag adds `src_res` and `dst_res` elements to the Tap
output, as described in #1170. These use the metadata labels in the tap
event to describe what Kubernetes resource the source and destination
peers belong to, based on what resource type is being tapped, and fall
back to pods if either peer is not a member of the specified resource
type.
In addition, when the resource type is not `namespace`, `src_ns` and
`dst_ns` elements are added, which show what namespaces the the source
and destination peers are in. For peers which are not in the Kubernetes
cluster, none of these labels are displayed.
The source metadata added in #1434 is used to populate the `src_res` and
`src_ns` fields.
Also, this branch includes some refactoring to how tap output is
formatted.
Signed-off-by: Eliza Weisman <eliza@buoyant.io>
* Upgrade to dep 0.5.0, go 1.10.3
* Remove existing dep binary if it's the wrong version
* Add version in filename of dep binary to prevent version conflicts
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
This an initial implementation of the `linkerd top` command. This command launches an ncurses style tabular view of current requests (using data from tap). Most of the command line arguments are the same as tap and allow selecting the resource to inspect and filtering which requests to view.
Fixes#1283
Signed-off-by: Alex Leong <alex@buoyant.io>
Based on @adleong's suggestion in
https://github.com/linkerd/linkerd2/pull/1434#pullrequestreview-145428857,
this branch adds label hydration from destination IPs to the Tap server.
This works the same as the label hydration for destination IPs added in
#1434. However, it is only applied to the destination fields of events
recorded by proxies in the inbound direction, since outbound
destinations are already labeled with metadata provided by the
Destination service.
This means that when a user taps inbound traffic, the CLI will show k8s
metadata labels for the destination peer (if it's available). This can
be useful especially when tapping several pods at once, as it makes it
easier to distinguish what pod received a request.
This branch also refactors how the label hydration is performed,
primarily to make adding it to the destination field less repetitive.
Also, the `hydrateIPLabels` function now mutates the label map in the
`TapEvent`, rather than returning a new map of labels, so that the case
where no pod was found doesn't require an additional allocation of an
empty map.
Signed-off-by: Eliza Weisman <eliza@buoyant.io>
The `TapEvent` protobuf contains two maps, `DestinationMeta` and
`SourceMeta`. The `DestinationMeta` contains all the metadata provided
by the proxy that originated the event (ultimately originating from the
Destination service), while the `SourceMeta` currently only contains the
source connection's TLS status.
This branch modifies the Tap server to hydrate the same set of metadata
from the source IP address, when the source was within the cluster. It
does this by adding an indexer of pod IPs to pods to its k8s API client,
and looking up IPs against this index. If a pod was found, the extra
metadata is added to the tap event sent to the client.
This branch also changes the client so that if a source pod name was
provided in the metadata, it prints the pod name rather than the IP
address for the `src` field in its output. This mimics what is currently
done for the `dst` field in tap output. Furthermore, the added source
metadata will be necessary for adding src resource types to tap output
(see issue #1170).
Signed-off-by: Eliza Weisman <eliza@buoyant.io>
Fixes#1405.
According to the Kubernetes Endpoints API documentation, the `name`
field in the `EndpointPort` response object is "Optional if only one
port is defined". (see
https://v1-9.docs.kubernetes.io/docs/reference/generated/kubernetes-api/v1.9/#endpointport-v1-core)
However, when the Destination service an endpoints response for a
service with a named target port, it expects the ports in the endpoints
response to have the same name as the target port in the service.
When a user creates a `NodePort` service with an unnamed port that
targets a named container port, this behaviour results in Linkerd
failing to route to that service by hostname. Without Linkerd injected,
the hostname is still reachable.
This branch fixes this issue by changing the `endpointsToAddresses`
function in `endpoints_watcher.go` to handle the case when an endpoints
response contains only a single unnamed port.
I've manually verified that this fixes the issue described in #1405.
Signed-off-by: Eliza Weisman <eliza@buoyant.io>
The `reader.Read` method only reads as many bytes as are currently available from reader. When reading the 4 byte message length header, if not all 4 of those bytes are available, `Read` will only read the available bytes and return. This causes alignment issues when the message body is read and there are still unread header bytes in the reader. These bytes will appear at the beginning of the message body and cause a crash when the message is unmarshalled.
Use `io.ReadFull` to ensure that we read all 4 of the message length header bytes.
Fixes#1287
Signed-off-by: Alex Leong <alex@buoyant.io>
* Update ant to 3.7.2
* Add autocomplete of namespaces/resources to Tap in web ui
* Add form fields for authority/path/method/rps/scheme
* Add the ability to clear error messages to the error banner
* Add error listener to ws object
Speed up incremental rebuilds by avoiding relinking the controller
and/or web executables when changes are made to unrelated files.
Before this change, any time the git tag changed, the executables
would have to be rebuilt (relinked at least), even if no Go files
changed.
Validated by running the integration tests.
Signed-off-by: Brian Smith <brian@briansmith.org>
Adds a tap endpoint in the web api that communicates with the dashboard
via websockets.
I've moved a bunch of code from the cli tap.go into utils so that the code
can be shared between web and CLI. I think we should consider making the
display more suited to web, but in the short term, reusing the CLI's
rendering of tap events works.
Adds a Tap page in the Web UI that you can use to make tap requests.
The form currently only allows you to enter a resource and namespace,
other filters coming in a follow-up branch.
`ca-bundle-distributor` described the original role of the program but
`ca` ("Certificate Authority") better describes its current role.
Signed-off-by: Brian Smith <brian@briansmith.org>
* Stop using `installsuffix` when building Go code.
See https://plus.google.com/117192131596509381660/posts/eNnNePihYnK.
`-installsuffix cgo` isn't necessary as of Go 1.10 (where build caching
changed substantially) and it probably wasn't necessary earlier.
Signed-off-by: Brian Smith <brian@briansmith.org>
* Ensure destination service always sends pod metadata
* Fix test that relied on hash ordering
* Stop using protobuf structs as map keys, fix logging
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
This PR begins to migrate Conduit to Linkerd2:
* The proxy has been completely removed from this repo, and is now located at
github.com/linkerd/linkerd2-proxy.
* A `Dockerfile-proxy` has been added to fetch the most-recently published proxy
binary from build.l5d.io.
* Proxy-specific protobuf bindings have been moved to
github.com/linkerd/linkerd2-proxy-api.
* All docker images now use the gcr.io/linkerd-io registry.
* `inject` now uses `LINKERD2_PROXY_` environment variables
* Go paths have been updated to reflect the new (future) repo location.
* Fix bug where we were using dst_authorities as a group by instead of authorities
* Add test to make sure we don't dst_authorities
Previously, we were only checking to make sure we didn't add
dst_authorities in the query labels in promDstQueryLabels but we
weren't checking the groupBy labels in promDstGroupByLabelNames -
this caused us to try to query for dst_authorities when a --from
query was sent. There are no dst_authorities, so there would be no
named results.
- Add Reason to the error data passed from the api
- Rewrite error logic in the UI to try to make it clearer
- Show 0/0 pods meshed instead of 0/0 pods meshed (N/A) if 0 pods are meshed
Create a ephemeral, in-memory TLS certificate authority and integrate it into the certificate distributor.
Remove the re-creation of deleted ConfigMaps; this will be added back later in #1248.
Signed-off-by: Brian Smith brian@briansmith.org
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
The proxy's metrics are instrumented with a `tls` label that describes
the state of TLS for each connection and associated messges.
This same level of detail is useful to get in `tap` output as well.
This change updates Tap in the following ways:
* `TapEvent` protobuf updated:
* Added `source_meta` field including source labels
* `proxy_direction` enum indicates which proxy server was used.
* The proxy adds a `tls` label to both source and destination meta indicating the state of each peer's connection
* The CLI uses the `proxy_direction` field to determine which `tls` label should be rendered.
I realized that our stat summary expectation checker would only check the actual
proto responses against the expectations if the expectations were non-empty.
Problem
If we expected empty results and the api returned actual results, we never actually
check those results against the expectations.
The bug can be reproduced by replacing any nonzero metric we expect in
expectedResponse with expectedResponse: genEmptyResponse()
The tests on master will still pass.
Solution
Remove this line and ensure we get the expected number of stat tables.
- Return pod uptimes from the GetPods endpoint
- Adds filtering by namespace to api.GetPods
- Adds a --namespace filter to conduit get pods
- Adds pod uptimes to the controller component toolitps on the ServiceMesh page
- Moves the ServiceMesh page back to using /api/pods
Adds the ability to query by a new non-kubernetes resource type, "authorities",
in the StatSummary api.
This includes an extensive refactor of stat_summary.go to deal with non-kubernetes
resource types.
- Add documentation to Resource in the public api so we can use it for authority
- Handle non-k8s resource requests in the StatSummary endpoint
- Rewrite stat summary fetching and parsing to handle non-k8s resources
- keys stat summary metric handling by Resource instead of a generated string
- Adds authority to the CLI
- Adds /authorities to the Web UI
- Adds some more stat integration and unit tests
* Add TLS support to `conduit inject`.
Add the settings needed to enable TLs when `--tls=optional` is passed on the
commend line. Later the requirement to add `--tls` will be removed.
Signed-off-by: Brian Smith <brian@briansmith.org>
* Update dest service with a different tls identity strategy
* Send controller namespace as separate field
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
* Add CA certificate bundle distributor to conduit install
* Update ca-distributor to use shared informers
* Only install CA distributor when --enable-tls flag is set
* Only copy CA bundle into namespaces where inject pods have the same controller
* Update API config to only watch pods and configmaps
* Address review feedback
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
* Add controller admin servers and readiness probes
* Tweak readiness probes to be more sane
* Refactor based on review feedback
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
Don't allow the CLI or Web UI to request named resources if --all-namespaces is used.
This follows kubectl, which also does not allow requesting named resources
over all namespaces.
This PR also updates the Web API's behaviour to be in line with the CLI's.
Both will now default to the default namespace if no namespace is specified.
Problem
`conduit stat` would cause a panic for any resource that wasn't in the list
of StatAllResourceTypes
This bug was introduced by https://github.com/runconduit/conduit/pull/1088/files
Solution
Fix writeStatsToBuffer to not depend on what resources are in StatAllResourceTypes
Also adds a unit test and integration test for `conduit stat ns`
* dest service: close open streams on shutdown
* Log instead of print in pkg packages
* Convert ServerClose to a receive-only channel
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
- It would be nice to display container errors in the UI. This PR gets the pod's container
statuses and returns them in the public api
- Also add a terminationMessagePolicy to conduit's inject so that we can capture the
proxy's error messages if it terminates
protobuf has a `go_package` option that can be used to explicitly name
Go packages such that they can be imported without additional rewrites.
This allows us to store proto files without additional, redundant
directories (which were used for packaging hints, previously).
This change adds an explicit `go_package` to all .proto files and
updates `bin/protoc-go.sh` to ensure these packages are output into
$GOPATH (so that the go_package can be absolute). This removes the need
to manually rewrite imports in bin/protoc-go.sh.
* Update destination service ot use shared informer instead of custom endpoints informer
* Add additional tests for dst svc endpoints watcher
* Remove service ports when all listeners unsubscribed
* Update go deps
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
* Update desintation service to use shared informer instead of pod watcher
* Add const for pod IP index name
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
Previously, in conduit stat all we would just print the map of stat results, which
resulted in the order in which stats were displayed varying between prints.
Fix:
Define an array, k8s.StatAllResourceTypes and use the order in this array to print
the map; ensuring a consistent print order every time the command is run.
Both the conduit stat command and web UI are showing failed and completed pods.
This change filters out those pods before returning the result to the client.
Fixes#1010
Signed-off-by: Ivan Sim <ihcsim@gmail.com>
Required for #1008.
This PR adds the `TlsIdentity` message to the Destination service proto,
to describe what strategy the proxy should use for verifying an endpoint's TLS
certificates. It also adds a `TlsIdentity` field to the `WeightedAddr` message.
Currently, there is one possible variant for `TlsIdentity`, `KubernetesPodName`,
which consists of the Kubernetes pod name of the endpoint, the namespace of
the endpoint, and the namespace of that pod's Conduit control plane. The proxy
should attempt to connect over TLS if the control plane namespace matches its
own control plane namespace. The pod name and namespace are used to verify
the endpoint's TLS certificate.
See https://github.com/runconduit/conduit/issues/386#issuecomment-392948046.
This change was initially part of #1008, but I factored it out to make the diff
smaller.
Signed-off-by: Eliza Weisman <eliza@buoyant.io>
- Update the `response_total` prometheus query of the StatSummary endpoint to also
break queries out by a `meshed` label.
- Add a 'Secured' column to the web UI/CLI stat displays, which indicate the percentage of traffic
starting and ending in the mesh
This meshed label is used in the CLI/Web UI to display a column of the percentage of traffic that
starts/ends in the mesh. (Which is a proxy indicator for whether that traffic is 'secured' when we
add TLS by default for intra mesh requests).
The `meshed` label is not yet added anywhere, so until it is supplied by the proxy, all traffic will
show up as 0% secured in the web/CLI.
The proxy does not yet support a `meshed` label.
In anticipation of a `meshed` label in the proxy, introduce this label
in `simulate-proxy`, for testing.
Relates to #306 and #386.
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
secured -> meshed
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
The StatSummary endpoint was dereferencing
StatSummaryRequest.Selector.Resource, causing a panic when it received
an empty request.
Fix StatSummary to use the nil-friendly
StatSummaryRequest.GetSelector().GetResource() methods, and add a test
to validate.
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Fix bug where we were dropping parts of the StatSummaryRequest
* Add tests for prometheus query strings and for failed cases
Problem
In #928 I rewrote the stat api to handle 'all' as a resource type. To query for all resource types,
we would copy the Resource, LabelSelector and TimeWindow of the original request, and then
go through all the resource types and set Resource.Type for each resource we wanted to get.
The bug is that while we copy over some fields of the original request, we didn't copy over all
of them - namely Resource.Name and the Outbound resource. So the Stat endpoint would
ignore any --to or --from flags, and would ignore requests for a specific named resource.
Solution
Copy over all fields from the request.
I've also added tests for this case. In this process I've refactored the stat_summary_test code
to make it a bit easier to read/use.
Allow the Stat endpoint in the public-api to accept requests for resourceType "all".
Currently, this queries Pods, Deployments, RCs and Services, but can be modified
to query other resources as well.
Both the CLI and web endpoints now work if you set resourceType to all.
e.g. `conduit stat all`
The way that git-related version information is linked into go binaries
busts Docker's cache such that every commit causes all binaries to
rebuilt.
In order to ameliorate this, we can build each binary once without
version information first so that its artifacts are cached. When Go
sources are not changed and only the version information changes, builds
are 4.3x faster than before (from 5+ minutes to <90s).
On `master`
Branch off of master and build (mostly cached):
```
:; time DOCKER_TRACE=1 bin/docker-build
...
DOCKER_TRACE=1 bin/docker-build 9.10s user 6.30s system 5% cpu 4:26.47 total
```
Rebuild without changing anything (highly cached):
```
:; time DOCKER_TRACE=1 bin/docker-build
...
DOCKER_TRACE=1 bin/docker-build 9.23s user 6.04s system 47% cpu 32.017 total
```
Update only the git sha and rebuild:
```
:; git ci -am 'bump it' --allow-empty
[ver/eg 2749eb3] bump it
:; time DOCKER_TRACE=1 bin/docker-build
...
DOCKER_TRACE=1 bin/docker-build 8.55s user 6.08s system 4% cpu 5:22.25 total
```
On this branch:
Rebuild without changing anything (highly cached):
```
:; time DOCKER_TRACE=1 bin/docker-build
...
DOCKER_TRACE=1 bin/docker-build 8.94s user 5.97s system 46% cpu 32.257 total
```
Update only the git sha and rebuild:
```
:; git ci -am 'bump it' --allow-empty
[ver/go-docker-cache-versionless 77a80b5] bump it
:; time DOCKER_TRACE=1 bin/docker-build
...
DOCKER_TRACE=1 bin/docker-build-cli-bin 2.02s user 1.34s system 9% cpu 34.144 total
```
* Fix bug where GetPodsFor(pod) was returning all pods in a namespace
Problem
In lister.GetPodsFor, when the input object was a pod, we would return all the pods in the namespace. I would expect GetPodsFor(pod) to return only one pod - the pod itself.
Cause
The cause of this is that when the object type was pod we were setting the selector to selector = labels.Everything() which gets all the pods in the namespace.
Fix
Special case GetPodsFor(pod) to return the pod itself, rather than looking up pods via labels.
* Modify the Stat endpoint to also return the count of failed pods
* Add comments explaining pod count stats
* Rename total pod count to running pod count
This is to support the service mesh overview page, as I'd like to include an indicator of
failed pods there.
After this was implemented we found that ExternalName services are
represented in DNS as CNAMEs, which means that the proxy's DNS
fallback logic can be used instead of doing DNS in the control
plane. Besides simplifying the controller, this will also increase
fidelity with the proxied pods' DNS configuration (improve
transparency).
Signed-off-by: Brian Smith <brian@briansmith.org>
The `conduit tap` command is now deprecated.
Replace `conduit tap` with `connduit tapByResource`. Rename tapByResource
to tap. The underlying protobuf for tap remains, the tap gRPC endpoint now
returns Unimplemented.
Fixes#804
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
public-api and and tap were both using their own implementations of
the Kubernetes Informer/Lister APIs.
This change factors out all Informer/Lister usage into the Lister
module. This also introduces a new `Lister.GetObjects` method.
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
The Kubernetes client-go Informer/Lister APIs are implemented in several
parts of the code base.
This change introduces a Lister module, providing Informer/Lister
capability through a simple interface. Once this merges, we can follow
up with moving public-api and tap onto Lister.
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
The TapByResource endpoint was previously a stub.
Implement end-to-end tapByResource functionality, with support for
specifying any kubernetes resource(s) as target and destination.
Fixes#803, #49
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
This PR removes the unused `request_duration_ms` and `response_duration_ms` histogram metrics from the proxy. It also removes them from the `simulate-proxy` script's output, and from `docs/proxy-metrics.md`
Closes#821
The Tap command leveraged new cli parsing code, enabling Kubernetes
resources specified as `(TYPE [NAME] | TYPE/NAME)`. The Stat command
did not use this.
Modify the Stat command to use the same cli flag parsing code as Tap.
Remove the to/from-resource flags from Stat.
Fixes#792
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
This PR makes two changes to the `simulate-proxy` script:
1. Removed the `protocol={"http", "tcp"}` label from TCP metrics. The proxy no longer adds this label (see https://github.com/runconduit/conduit/pull/785#discussion_r182563499).
2. Fixed failed responses being labeled with `classification="fail"` rather than `classification="failure"` (the label the proxy sets). I noticed that while I was here and decided to fix it as well.
Note that the first change required some minor changes to the `proxyMetricCollectors` struct in `simulate-proxy`; since the label cardinality for TCP open stats decreased by one due to removing the `protocol` label, it's no longer necessary for that struct to `haveCounterVec`/`GaugeVec` pointers for these stats. It now owns the actual `Counter`/`Gauge` instead. This means that the metric vecs that are created to be labeled for `inbound` and `outbound` are now stored as variables in the `newSimulatedProxy` function rather than going in a `proxyMetricCollectors` struct first. This shouldn't impact behaviour at all.
The `stat` command did not support `service` as a resource type.
This change adds `service` support to the `stat` command. Specifically:
- as a destination resource on `--to` commands
- as a target resource on `--from` commands
Fixes#805
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
This PR adds the transport-level metrics described in #742 to the `simulate-proxy` script. This will be useful while adding these metrics to the Grafana dashboard and/or CLI.
Closes#793
The existing `tap` command is being deprecated.
Introduce a `tapByResource` cli command. It supports tapping a Kubernetes
resource or collection of resources, optionally filtered by outbound resources.
This command will eventually replace `tap`.
Part of #778
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
This changes the public api to have a new rpc type, `TapByResource`.
This api supersedes the Tap api. `TapByResource` is richer, more closely
reflecting the proxy's capabilities.
The proxy's Tap api is extended to select over destination labels,
corresponding with those returned by the Destination api.
Now both `Tap` and `TapByResource`'s responses may include destination
labels.
This change avoids breaking backwards compatibility by:
* introducing the new `TapByResource` rpc type, opting not to change Tap
* extending the proxy's Match type with a new, optional, `destination_label` field.
* `TapEvent` is extended with a new, optional, `destination_meta`.
The Destination service does not provide ReplicaSet information to the
proxy.
The `pod-template-hash` label approximates selecting over all pods in a
ReplicaSet or ReplicationController. Modify the Destination service to
provide this label to the proxy.
Relates to #508 and #741
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Expose pod stats in CLI, web UI, and Grafana
* Fix js api helpers test
* Add outbound traffic stats to pod dashboard
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
When the Destination sees an IP address, it looks up Pods by that IP,
and associates Pod label data to it. If the lookup by IP returned more
than one Pod, it simply picked the first one. This is not correct,
specifically in cases where one pod is in a Running state, and others
are not.
Modify the Destination service to only return label data for Pods in the
Running state.
Fixes#773
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
The public-api previously only permitted 4 hard-coded time windows:
10s, 1m, 10m, 1h. This was primarily a relic of the recently removed
telemetry system.
Modify the public-api to validate the time string, but allow for any
window size, which is then passed through to Prometheus.
Fixes#686
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Add namespace as a resource type in public-api
The cli and public-api only supported deployments as a resource type.
This change adds support for namespace as a resource type in the cli and
public-api. This also change includes:
- cli statsummary now prints `-`'s when objects are not in the mesh
- cli statsummary prints `No resources found.` when applicable
- removed `out-` from cli statsummary flags, and analagous proto changes
- switched public-api to use native prometheus label types
- misc error handling and logging fixes
Part of #627
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Refactor filter and groupby label formulation
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
* Rename stat_summary.go to stat.go in cli
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
* Update rbac privileges for namespace stats
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
Conduit was relying on apps/v1 to Deployment and ReplicaSet APIs.
apps/v1 is not available on Kubernetes 1.8. This prevented the
public-api from starting.
Switch Conduit to use apps/v1beta2. Also increase the Kubernetes API
cache sync timeout from 10 to 60 seconds, as it was taking 11 seconds on
a test cluster.
Fixes#761
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Remove the telemetry service
The telemetry service is no longer needed, now that prometheus scrapes
metrics directly from proxies, and the public-api talks directly to
prometheus. In this branch I'm removing the service itself as well as
all of the telemetry protobuf, and updating the conduit install command
to no longer install the service. I'm also removing the old version of
the stat command, which required the telemetry service, and renaming the
statsummary command to stat.
* Fix time window tests
* Remove deprecated controller scrape config
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
The Prometheus client sometimes returns NaN if a calculation is invalid,
such as histogram_quantile when no requests have occurred.
Add IsNaN check in the public-api and set output to zero.
Fixes#747
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
The ListPods endpoint's logic resides in the telemetry service, which is
going away.
Move ListPods logic into public-api, use new k8s informer APIs.
Fixes#694
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
The new StatSummary endpoint was only providing request volume and
successs rate information.
Add support for retrieving latency stats via StatSummary. Also make
all prometheus calls in parallel, and implement kubernetes test
fixtures.
Fixes#681
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Switch public API to use cached k8s resources
* Move shared informer code to separate goroutine
* Fix spelling issue
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
The success rate calculation relies on the `classification` label, but
was incorrectly specifying `fail` rather than `failure`.
Fix public api to specify `failure`. Also re-org public api tests for
easier Kubernetes and Prometheus mocking.
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
The StatSummary logic was implemented as a method on http_server.
Move the StatSummary logic into grpc_server, for consistency with the
other endpoints.
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
The new statsummary command accepted friendly k8s names, which worked
for k8s queries, but Prometheus requires a specific key.
Modify the statsummary query to map friendly k8s names to canonical k8s
names when constructing the query. Then during the query, map the
canonical k8s name to a specific Prometheus label.
Fixes#695
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
Start implementing new conduit stat summary endpoint.
Changes the public-api to call prometheus directly instead of the
telemetry service. Wired through to `api/stat` on the web server,
as well as `conduit statsummary` on the CLI. Works for deployments only.
Current implementation just retrieves requests and mesh/total pod count
(so latency stats are always 0).
Uses API defined in #663
Example queries the stat endpoint will eventually satisfy in #627
This branch includes commits from @klingerf
* run ./bin/dep ensure
* run ./bin/update-go-deps-shas
The Destination service used slightly different labels than the
telemetry pipeline expected, specifically, prefixed with `k8s_*`.
Make all Prometheus labels consistent by dropping `k8s_*`. Also rename
`pod_name` to `pod` for consistency with `deployement`, etc. Also update
and reorganize `proxy-metrics.md` to reflect new labelling.
Fixes#655
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Define a new telemetry Stat API
Proposal definition for a new Stat API, for the purposes of satisfying the queries proposed in #627.
StatSummary will replace Stat once implemented and the original Stat deleted.
* Extracted logic from destination server
* Make tests follow style used elsewhere in the code
* Extract single interface for resolvers
* Add tests for k8s and ipv4 resolvers
* Fix small usability issues
* Update dep
* Act on feedback
* Add pod-based metric_labels to destinations response
* Add documentation on running control plane to BUILD.md
Signed-off-by: Phil Calcado <phil@buoyant.io>
* Fix mock controller in proxy tests (#656)
Signed-off-by: Eliza Weisman <eliza@buoyant.io>
* Address review feedback
* Rename files in the destination package
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
simulate-proxy uses a deployment object from kubernetes to simulate
each proxy metrics endpoint.
Modify simulate-proxy to instead use a pod to simulate each proxy
metrics endpoint. This ensures that each metrics endpoint consistently
represents a pod in kubernetes, including it's namespace, deployment,
and label information.
This change also adds support for:
- a new `metric-ports` flag, default is `10000-10009`.
- `classification`, `pod_name`, and `pod_template_hash` labels
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Extracted logic from destination server
* Make tests follow style used elsewhere in the code
* Extract single interface for resolvers
* Add tests for k8s and ipv4 resolvers
* Fix small usability issues
* Update dep
* Act on feedback
Signed-off-by: Phil Calcado <phil@buoyant.io>
simulate-proxy increments a single set of metrics on each iteration, and
also randomizes http status codes, leaving counters unchanged across
several collections.
Modify simuilate-proxy to increment all metrics on each iteration,
provide a 90% success rate, ensure a pod does not call itself, and
increase proxy count from 3 to 10.
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
* Add tests/utils/scripts for running integration tests
Add a suite of integration tests in the `test/` directory, as well as
utilities for testing in the `testutil/` directory.
You can use the `bin/test-run` script to run the full suite of tests,
and the `bin/test-cleanup` script to cleanup after the tests.
The test/README.md file has more information about running tests.
@pcalcado, @franziskagoltz, and @rmars also contributed to this change.
* Create TEST.md file at the root of the repo
* Update based on review feedback
* Relax external service IP timeout for GKE
* Update TEST.md with more info about different types of test runs
* More updates to TEST.md based on review feedback
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
The Prometheus scrape config collects from Conduit proxies, and maps
Kubernetes labels to Prometheus labels, appending "k8s_".
This change keeps the resultant Prometheus labels consistent with their
source Kubernetes labels. For example: "deployment" and
"pod_template_hash".
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
Have the controller tell the client whether the service exists, not
just what are available. This way we can implement fallback logic to
alternate service discovery mechanisms for ambigious names.
Signed-off-by: Brian Smith <brian@briansmith.org>
Signed-off-by: Kevin Lingerfelt <kl@buoyant.io>
The Prometheus config in the docker-compose environment had fallen
behind the prod setup.
This change updates the docker-compose environment in the following
ways:
- Prometheus config more closely matches prod, based on #583
- simulate-proxy labels matches prod, based on #605
- add Grafana container
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
The simulate-proxy script pushes metrics to the telemetry service. This PR modifies the script to expose metrics to a prometheus endpoint. This functionality creates a server that randomly generates response_total, request_totals, response_duration_ms and response_latency_ms. The server reads pod information from a k8s cluster and picks a random namespace to use for all exposed metrics.
Tested out these changes with a locally running prometheus server. I also ran the docker-compose.yml to make sure metrics were being recorded by the prometheus docker container.
fixes#498
Signed-off-by: Dennis Adjei-Baah <dennis@buoyant.io>
* Most controller listeners should only bind on localhost
* Use default listening addresses in controller components
* Review feedback
* Revert test_helper change
* Revert use of absolute domains
Signed-off-by: Alex Leong <alex@buoyant.io>
Shortly after conduit is installed in k8s environment. The control plane component that establishes a watch endpoint with k8s run in to networking issues during proxy initialization. During failure, each watcher fails to retry its connection to k8s watch endpoint which leads to timeouts and eventually, multiple controller pod restarts.
This PR adds retry logic to each "watch" enabled package.
fixes#478
Signed-off-by: Dennis Adjei-Baah <dennis@buoyant.io>
When the conduit proxy is injected into the controller pod, we observe controller pod proxy stats show up as an "outbound" deployment for an unrelated upstream deployment. This may cause confusion when monitoring deployments in the service mesh.
This PR filters out this "misleading" stat in the public api whenever the dashboard requests metric information for a specific deployment.
* exclude telemetry generated by the control plane when requesting deployment metrics
fixes#370
Signed-off-by: Dennis Adjei-Baah <dennis@buoyant.io>
The dns_test had assumed DNS changes were deterministically ordered, but
util.DiffAddresses uses a map and therefore does not guarantee ordering.
Fix dns_test to sort TCP Addresses prior to comparison.
Fixes#515
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
When attempting to tap N pods when N is greater than the target rps, a rounding error occurs that requests 0 rps from each pod and no tap data is returned.
Ensure that tap requests at least 1 rps from each target pod.
Tested in Kubernetes on docker-for-desktop with a 15 replica deployment and a maxRps of 10.
Signed-off-by: Alex Leong <alex@buoyant.io>
The control plane is proxied through the Conduit proxy. The Conduit
proxy is based on the base image, and the control plane containers
and the proxy share a networking namespace. This means we don't
need the extra base utilities in the controller images since we can
use the utilties in the proxy image.
This is a step towards building the initial no-networking Conduit CA
pod. Since the Conduit CA will not do any networking of its own, we
networking debugging utilties are not helpful for it. They are
actually an unnecessary risk because they could facilitate the
exfiltration of the private key of the CA. (The Conduit CA pod won't
have the Conduit Proxy injected into it either.)
This also simplifies & slightly speeds up the building of the
controller images. This is a stepping stone towards being able to
build the controller images without `docker build` to improve build
times.
Signed-off-by: Brian Smith <brian@briansmith.org>
* Use Go 1.10.0 to build Go components.
Take advantage of the new build cache in Go 1.10. Future work on improving
build performance will utilize the build cache further.
Signed-off-by: Brian Smith <brian@briansmith.org>
Previously Dockerfile-go-deps was converted from a multi-stage Dockefile
to a single-stage Dockerfile in anticipation of enabling efficient use
of `--cache-from` in CI. However, that resulted in the image ballooning
in size because it contained the Git repo for every package downloaded
by `dep ensure`.
Bring the image back down to the proper size by removing the temporary
files created.
Signed-off-by: Brian Smith <brian@briansmith.org>
The instance cache that powers the ListPods API is stored in memory in the telemetry service. This means that when there are multiple replicas of the telemetry service, each replica will have a distinct, incomplete view of the added pods based on which pods report to that telemetry replica. This causes the data plane bubbles on the dashboard to not all be filled in, and to flicker with each data refresh.
We create a Prometheus counter called reports_total which has pod as a label. Whenever a telemetry service instance receives a report from a pod, it increments reports_total for that pod. This allows us to remove the in-memory instance cache and instead query Prometheus to see if each pod has had a report in the last 30 seconds.
Fixes#337
Signed-off-by: Alex Leong <alex@buoyant.io>
In PR #298 we moved time window parsing (10s => (time.now - 10s,
time.now) down the stack to immediately before the query. This had the
unintended effect of creating parallel latency quantile requests with
slightly different timestamps.
This change parses the time window prior to latency quantile fan out,
ensuring all requests have the same timestamp.
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
PR #298 moved summary (non-timeseries) requests to Prometheus' Query
endpoint, with no timestamp provided. This Query endpoint returns a
single data point with whatever timestamp was provided in the request.
In the absense of a timestamp, it uses current server time. This causes
the Public API to return discreet data points with slightly different
timestamps, which is unexpected behavior.
Modify the Public API -> Telemetry -> Prometheus request path to always
require a timestamp for single data point requests.
Fixes#340
Signed-off-by: Andrew Seigner <siggy@buoyant.io>
On my system (i9-7960x running Docker natively in Linux) this regularly saves
over 11 seconds of build time when a file under pkg/ changes and over 1.5
seconds of build time when a file under controller/ changes. Since most
contributors are running Docker in a VM on less powerful computers, the
savings for most contributors should be significantly greater.
I imagine the savings for web/ and cli/ and proxy-init/ are similar, but I
did not measure them.
Signed-off-by: Brian Smith <brian@briansmith.org>
Precompiling pkg/ in an earlier layer saves ~10 seconds of wall clock
time on an incremental build on my machine (i9-7960x) when I update a
file in controller/ such as controller/destination/server.go. This
makes a significant difference in the edit-build-test loop.
Signed-off-by: Brian Smith <brian@briansmith.org>
Eliza Weisman
Alex Leong
zak
Alena Varkockova
Dennis Adjei-Baah
Andrew Seigner
Kevin Lingerfelt
Radu M
Cody Vandermyn
Alejandro Pedraza
Oliver Gould
Risha Mars
Kevin Leimkuhler
Igor Zibarev
Ivan Sim
Ben Lambert
Sean McArthur
Brian Smith
Phil Calçado
Dennis Adjei-Baah