Skip to content
This repository has been archived by the owner on Jun 28, 2022. It is now read-only.

Update Helm release cert-manager to v1.8.2 #5

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Update Helm release cert-manager to v1.8.2 #5

wants to merge 1 commit into from

Conversation

renovate[bot]
Copy link

@renovate renovate bot commented May 24, 2022 •
edited
Loading

Mend Renovate

This PR contains the following updates:

Package Type Update Change
cert-manager helm_release minor v1.1.0 -> v1.8.2

Release Notes

jetstack/cert-manager

v1.8.2

Compare Source

cert-manager is the easiest way to automatically manage certificates in Kubernetes and OpenShift clusters.

v1.8.2 is in effect a bug fix release which increases some hard-coded timeouts which were preventing the use of certain ACME issuers
which sometimes had slower response times. This is known to include ZeroSSL and Sectigo.

These issues were reported by many different users and We'd like to thank the following for their help, suggestions and feedback on this topic:

Thanks also to the cert-manager maintainers who were involved in reviewing this fix and helping to move things forwards:

Changes since v1.8.1

Bug
  • Increase timeouts for issuer and clusterissuer controllers to 2 minutes and increase ACME client HTTP timeouts to 90 seconds, in order to enable the use of slower ACME issuers which take a long time to process certain requests. (#​5231, @​JoooostB @​SgtCoDFish)
Other (Cleanup)

v1.8.1

Compare Source

cert-manager is the easiest way to automatically manage certificates in Kubernetes and OpenShift clusters.

1.8.1 is a patch release rebuilding cert-manager 1.8 using the latest version of Go.

Changelog since cert-manager 1.7.1

  • Reverts a check for Prometheus APIs before creating cert-manager ServiceMonitors which broke users' GitOps flows (cert-manager#​5204)

  • Bumps the version of Go used to build the cert-manager binaries to 1.17.11 which fixes a few CVEs (we don't think that those were likely to be exploited in cert-manager) (cert-manager#​5203, @​irbekrm )

v1.8.0

Compare Source

cert-manager is the easiest way to automatically manage certificates in Kubernetes and OpenShift clusters.

cert-manager 1.8 includes wider support for Kubernetes server-side-apply, a new build and development experience based around
Makefiles rather than Bazel, and a range of other improvements, tweaks and bug fixes.

Version 1.8 also marks our first release in which the Go import path for cert-manager is that of the repo's new home:

github.com/cert-manager/cert-manager

Breaking Changes (You MUST read this before you upgrade!)
Validation of the rotationPolicy field

The field spec.privateKey.rotationPolicy on Certificate resources is now validated. Valid options are Never and Always. If you are using a GitOps flow and one of your YAML manifests contains a Certificate with an invalid value, you will need to update it with a valid value to prevent your GitOps tool from failing on the new validation. Please follow the instructions listed on the page Upgrading from v1.7 to v1.8. (#​4913, @​jahrlin)

What happens if I upgrade to 1.8.0 without doing the above steps?

After upgrading to 1.8.0, when updating existing Certificate objects that have an incorrect value for rotationPolicy, Kubernetes clients such as kubectl, Helm, or ArgoCD will start showing the following message:

Certificate.cert-manager.io "my-cert" is invalid: spec.privateKey.rotationPolicy: Unsupported value: "Foo": supported values: "Never", "Always".
Why was this change necessary?

Previously, when the value of the rotationPolicy field was set to an incorrect value, you would not know since no event or condition would be visible on the Certificate itself. The only way to know that something was wrong was to dig into the cert-manager-controller logs and see the message "Certificate with unknown certificate.spec.privateKey.rotationPolicy value":

I0329 12:43:13.325771       1 keymanager_controller.go:176] cert-manager/certificates-key-manager "msg"="Certificate with unknown certificate.spec.privateKey.rotationPolicy value" "key"="default/my-cert" "rotation_policy"="Foo"

This change was implemented in #​4913.

Changed Container Layouts

This only affects you if you're modifying cert-manager containers in some way, such as adding init scripts or otherwise
changing how the binaries inside the containers are called.

Bazel has a unique way of creating containers, which places the actual binary at a long unusual path. For the v1.7.0 cert-manager-webhook
container for example, the binary is placed at /app/cmd/webhook/webhook.runfiles/com_github_jetstack_cert_manager/cmd/webhook/webhook_/webhook
and /app/cmd/webhook/webhook is provided as a symlink to the binary.

This is simplified in our new build system; we only place a single binary at /app/cmd/webhook/webhook and the old path disappears.
This applies to all cert-manager containers.

We also removed the "LICENSES" file from the containers and replaced it with a link to the cert-manager repo.

.exe Extension on Windows

We package cmctl and kubectl_cert-manager for Windows on amd64 platforms, but previously the binaries had the
same names as the binaries on other platforms, e.g. cmctl with no file extension.

In 1.8.0 and later, the binaries now have a .exe extension since this is standard practice on Windows. This could affect you
if you're calling the binary in a Powershell script, for example.

We've also now added zip-compressed versions of the cmctl and kubectl_cert-manager binaries on Windows, since .tar.gz is less
common on Windows.

Changed Import Path

This will only affect you if you're writing code in Go which imports cert-manager as a module, which we generally recommend against
doing in most cases.

All versions of cert-manager prior to v1.8.0 used a Go import path corresponding to the old cert-manager repository, github.com/jetstack/cert-manager.

v1.8.0 marks the first release in which the import path changes to the new location, github.com/cert-manager/cert-manager.

We have a guide for Importing cert-manager in Go on cert-manager.io with all the details, including
details on why we don't recommend importing cert-manager as a module if that's avoidable.

Major Themes
Server-Side Apply

cert-manager v1.8.0 adds initial support for Kubernetes Server-Side Apply, which became stable
in Kubernetes 1.22. This support is behind a feature gate for now, and is only supported by cert-manager on Kubernetes 1.22 and later.

Server-Side Apply helps to ensure that changes to resources are made in a managed way, and aims to prevent certain classes of bugs. Notably, it should
eliminate conflicts when multiple controllers try to apply status changes to a single resource. You'll likely have seen messages relating to this kind of
conflict in logs before, e.g.:

I0119 12:34:56.000000       1 controller.go:161] cert-manager/controller/certificaterequests-issuer-acme "msg"="re-queuing item due to optimistic locking on resource" "key"="my-namespace/my-cr" "error"="Operation cannot be fulfilled on certificaterequests.cert-manager.io \"my-cr\": the object has been modified; please apply your changes to the latest version and try again"

These conflicts aren't usually actually a problem which will block the issuance of a certificate, but they can delay things as they cause extra
reconcile loops. Server-side apply cleans things up, which should mean less noise in logs and fewer pointless reconcile loops.

If you want to test it out, you can enable alpha-level cert-manager Server-Side Apply support through the
--feature-gates controller flag.

From Bazel to Make

A common theme when someone tries to make a change to cert-manager for the first time is that they ask for help with navigating Bazel, which cert-manager
used as its build tool. Helping people with Bazel isn't easy; it's an incredibly powerful tool, but that power also brings a lot of complications
which can seriously get in the way of being able to make even simple changes to the code base. Even developers who are familiar with contributing
to open source projects in Go can find it daunting to make changes thanks to Bazel.

The problem isn't limited to open-source contributors; many of cert-manager's maintainers also struggle with configuring and changing Bazel, too.

cert-manager 1.8 is the first release which is built and tested using a newly written make-based build system. We believe that this new build system should
make it much simpler to understand and change the commands which are being run behind the scenes to build and test cert-manager. In time, we'll fully
document the new build system, ensure it's at full feature-parity with Bazel and then remove all references to Bazel across the codebase.

A neat side effect of this change is that our build times have significantly improved. Bazel took around 14 minutes to build every cert-manager
artifact for every platform during a release, while the new make build system can do the same (and more) in under 5 minutes.

Exponential backoff after a failed issuance

cert-manager v1.8.0 introduces exponential backoff after failed certificate issuance.

Previously, a failed issuance was retried every hour which — especially in larger cert-manager installations — could cause rate limits to be hit as well as overwhelm external services. Failed attempts
are now retried with a binary exponential backoff starting with 1h then 2h, 4h up to a maximum of 32h. As part of the new backoff behavior, a new failedIssuanceAttempts field was added to the
Certificate spec to track the number of currently failed issuances.

The cmctl renew command command can still be used to force Certificate renewal immediately.

We're also considering reducing the initial backoff from 1 hour. If you have a use case where this would be useful please do comment on our tracking issue.

Community

cert-manager thrives thanks to the community and we're always grateful for receiving open-source contributions!

Thanks to the following community members who landed a commit in this release:

Thanks also to the cert-manager maintainer team involved with this release

Changelog since v1.7.0

Feature
  • ACTION REQUIRED: The field spec.privateKey.rotationPolicy on Certificate resources is now validated. Valid options are Never and Always. If you are using a GitOps flow and one of your YAML manifests contains a Certificate with an invalid value, you will need to update it with a valid value to prevent your GitOps tool from failing on the new validation. (#​4913, @​jahrlin)
  • Build: add make targets for running unit and integration tests, as part of the Bazel replacement. (#​4865, @​SgtCoDFish)
  • Build: add make targets for running the end-to-end tests, as part of the Bazel replacement. (#​4914, @​maelvls)
  • cert-manager now supports the field spec.expirationSeconds on Kubernetes CertificateSigningRequest resources. Using this field requires Kubernetes 1.22. You can still use the annotation experimental.cert-manager.io/request-duration to request a duration. (#​4957, @​enj)
  • cert-manager now properly updates the content of the data keys tls-combined.pem and key.der on Secret resources that are associated to Certificate resources that use the field additionalOutputFormats. The field additionalOutputFormat is an alpha feature and can be enabled by passing the flag --feature-gates=AdditionalCertificateOutputFormats=true to the cert-manager controller. (#​4813, @​JoshVanL)
  • ClusterRoles aggregation to user-facing admin/edit/view ClusterRoles can be optionally turned off (#​4937, @​illrill)
  • ACTION REQUIRED: Server-Side Apply: the feature gate ServerSideApply=true now configures the ingress-shim and gateway-shim controllers to use Kubernetes Server-Side Apply on Certificate resources. When upgrading to cert-manger 1.8 with ServerSideApply=true, do make sure there are no Challenge resources currently in the cluster. If there are some, you will need to manually delete them once they are in 'valid' state as cert-manager post-1.8 with the Server-Side Apply feature is not able to clean up Challenge resources created pre-1.8. (#​4811, @​JoshVanL)
  • Server-Side Apply: the feature gate ServerSideApply=true configures the certificaterequests-* controllers to use Kubernetes Server-Side Apply on CertificateRequest resources. (#​4792, @​JoshVanL)
  • Server-Side Apply: the feature gate ServerSideApply=true configures the certificates-* controllers to use Kubernetes Server-Side Apply on Certificate resources. (#​4777, @​JoshVanL)
  • Server-Side Apply: the feature gate ServerSideApply=true configures the CertificateSigningRequest controllers to use Kubernetes Server-Side Apply on CertificateSigningRequest resources. (#​4798, @​JoshVanL)
  • Server-Side Apply: the feature gate ServerSideApply=true configures the issuers and clusterissuers controllers to use Kubernetes Server-Side Apply on Issuer and ClusterIssuer resources. (#​4794, @​JoshVanL)
  • Server-Side Apply: the feature gate ServerSideApply=true configures the orders controller to use Kubernetes Server-Side Apply on Order resources. (#​4799, @​JoshVanL)
  • The annotation experimental.cert-manager.io/request-duration now has a minimum value of 600 seconds. This annotation This change ensures compatibility with the Kubernetes resource CertificateSigningRequest, which requires a minimum of 600 seconds on the field spec.expirationSeconds. (#​4973, @​irbekrm)
  • The annotation ingress.kubernetes.io/whitelist-source-range used by the Ingress shim when creating Ingress resources can now be overridden by setting the field ingressTemplate on the Issuer and ClusterIssuer. (#​4789, @​tasharnvb)
  • The experimental Gateway API support now uses the v1alpha2 CRDs. (#​4791, @​jakexks)
  • The user-agent used by cert-manager in its Kubernetes API clients and ACME clients now takes the form cert-manager-<component name>/<version> (<os>/<arch>) cert-manager/<git commit>. Another change is the addition of specific field managers strings; previously, all the controllers had the same field manager cert-manager. Now, each controller has its own field manager string of the form cert-manager-<controller name>. (#​4773, @​JoshVanL)
  • You can now uninstall cert-manager using the command cmctl experimental uninstall. (#​4897, @​jahrlin)
  • You can now use an external issuer resource as the default issuer when using the Ingress shim feature. The default issuer can be set using the flags --default-issuer-group, --default-issuer-kind, and --default-issuer-name. (#​4833, @​jakexks)
Design
  • ACTION REQUIRED: The import path for cert-manager has been updated to github.com/cert-manager/cert-manager. If you import cert-manager as a go module (which isn't currently recommended), you'll need to update the module import path in your code to import cert-manager 1.8 or later. (#​4587, @​SgtCoDFish)
Bug or Regression
  • ACTION REQUIRED: The field additionalOutputFormats, which is available as an alpha feature on Certificate resources, is now correctly validated. Previously, it would only get validated when the privateKey field was set on the Certificate. If you are using the additionalOutputFormats field, you will want to add the feature gate AdditionalCertificateOutputFormats to both the webhook and the controller. Previously, you only needed to set AdditionalCertificateOutputFormats on the controller. If the feature gate is missing on either the controller or the webhook, you won't be able to use the additionalOutputFormat field. (#​4814, @​JoshVanL)
  • The Go version used to build the cert-manager binaries has been bumped to 1.17.8 to fix a slew of CVEs (none of which were likely to be exploited). (#​4970, @​vhosakot)
  • Helm: the default nodeSelector is now kubernetes.io/os: linux. If this label isn't present on any nodes in the cluster, the nodeSelector will need to be overwritten, or that label added to some nodes. (#​3605, @​mikebryant)
  • Use multivalue records instead of simple records for the AWS Route53 ACME DNS challenge solver, to allow for multiple challenges for the same domain at the same time (#​4793, @​fvlaicu)
Other (Cleanup or Flake)
  • Aggregated admin and edit roles will now include permissions to update certificates' status, which will allow namespace admins and editors to run the cmctl renew command in their namespaces. (#​4955, @​andreadecorte)
  • Cleanup: No longer log an error when cert-manager encounters a conflict in the secrets manager, in favor of always force applying. (#​4815, @​JoshVanL)
  • Failed certificate issuances are now retried with an exponential backoff where the backoff periods are 1h, 2h, 4h, 8h, 16h, 32h. A new field failedIssuanceAttempts is now set by cert-manager on the Certificate status. This field keeps track of consecutive failed issuances. The backoff period gets reset after a successful issuance. Like before, updating a field on a failed Certificate (such as spec.dnsNames) or running the command cmctl renew continues to trigger a re-issuance. (#​4772, @​irbekrm)
  • When starting up, cert-manager now solely relies on Lease objects to perform the leader election. Previously, cert-manager supported both ConfigMap and Lease objects for leader election. Existing ConfigMap resources used for leader election will remain and will need deleting manually. A side effect of this is that you cannot upgrade to v1.8.0 from cert-manager 1.3 (although upgrading multiple versions at a time was never supported). (#​4935, @​davidsbond)
  • Helm: you can now set custom labels on the ServiceAccount resources using the values serviceAccount.labels, cainjector.serviceAccount.labels, webhook.serviceAccount.labels, and startupapicheck.serviceAccount.labels. (#​4932, @​4molybdenum2)
Uncategorized
  • Introducing a new metric controller_sync_error_count counting the number of errors during sync() of a controller. (#​4987, @​jayme-github)
  • When creating an acmesolver pod, cert-manager now sets allowPrivilegeEscalation to false by default. The Helm chart now also sets securityContext.allowPrivilegeEscalation to false by default for the controller, cainjector, and webhook pods as well as for the startupapicheck job. (#​4953, @​ajvn)

v1.7.3

Compare Source

cert-manager is the easiest way to automatically manage certificates in Kubernetes and OpenShift clusters.

v1.7.3 is in effect a bug fix release which increases some hard-coded timeouts which were preventing the use of certain ACME issuers
which sometimes had slower response times. This is known to include ZeroSSL and Sectigo.

These issues were reported by many users. We'd like to thank the following for their help and feedback on this topic:

Thanks also to the cert-manager maintainers who were involved in reviewing this fix and helping to move things forwards:

Changes since v1.7.2
Bug
  • Increase timeouts for issuer and clusterissuer controllers to 2 minutes and increase ACME client HTTP timeouts to 90 seconds, in order to enable the use of slower ACME issuers which take a long time to process certain requests. (#​5232, @​JoooostB @​SgtCoDFish)
Other (Cleanup)

v1.7.2

Compare Source

cert-manager is the easiest way to automatically manage certificates in Kubernetes and OpenShift clusters.

1.7.2 is a minor release rebuilding cert-manager 1.7 using the latest version of Go. This eliminates a few security vulnerabilities which have accumulated in Go since the last release.

We don't believe any of those vulnerabilities were practically exploitable or relevant to cert-manager, but we decided to rebuild to keep up to date anyway.

Changelog since cert-manager 1.7.1

Bug or Regression

v1.7.1

Compare Source

cert-manager is the easiest way to automatically manage certificates in Kubernetes and OpenShift clusters.

Version 1.7.1 fixes a bug which was discovered in 1.7.0 relating to the new additionalOutputFormat feature.

Changelog since v1.7.0

Bug or Regression
  • Fix: The alpha feature Certificate's additionalOutputFormats is now correctly validated at admission time, and no longer only validated if the privateKey field of the Certificate is set. The Webhook component now contains a separate feature set.
    AdditionalCertificateOutputFormats feature gate (disabled by default) has been added to the webhook. This gate is required to be enabled on both the controller and webhook components in order to make use of the Certificate's additionalOutputFormat feature. (#​4816, @​JoshVanL)

v1.7.0

Compare Source

cert-manager is the easiest way to automatically manage certificates in Kubernetes and OpenShift clusters.

Version 1.7 brings new private key output formats, configuration improvements for the webhook, some long-awaited code cleanup, a fix for ingress class semantics and a bunch of other changes.

Breaking Changes (You MUST read this before you upgrade!)

Removal of Deprecated APIs

⚠ Following their deprecation in version 1.4, the cert-manager API versions v1alpha2, v1alpha3, and v1beta1 have been removed. You must ensure that all cert-manager custom resources are stored in etcd at version v1 and that all cert-manager CustomResourceDefinitions have only v1 as the stored version before upgrading.

Since release 1.7, cmctl can automatically migrate any deprecated API resources. Please [download cmctl-v1.7.0][download cmctl-v1.7.0] and read Migrating Deprecated API Resources for full instructions.

Ingress Class Semantics

In 1.7, we have reverted a change that caused a regression in the ACME Issuer. Before 1.5.4, the Ingress created by cert-manager while solving an HTTP-01 challenge contained the kubernetes.io/ingress.class annotation:

apiVersion: networking.k8s.io/v1beta1
kind: Ingress
metadata:
  annotations:
    kubernetes.io/ingress.class: istio # The `class` present on the Issuer.

After 1.5.4, the Ingress does not contain the annotation anymore. Instead, cert-manager uses the ingressClassName field:

apiVersion: networking.k8s.io/v1
kind: Ingress
spec:
  ingressClassName: istio # 🔥 Breaking change!

This broke many users that either don't use an Ingress controller that supports the field (such as ingress-gce and Azure AGIC), as well as people who did not need to create an IngressClass previously (such as with Istio and Traefik).

The regression is present in cert-manager 1.5.4, 1.6.0, 1.6.1. It is only present on Kubernetes 1.19+ and only appears when using an Issuer or ClusterIssuer with an ACME HTTP-01 solver configured.

In 1.7, we have restored the original behavior which is to use the annotation. We will also backport this fix to 1.5.5 and 1.6.2, allowing people to upgrade safely.

Most people won't have any trouble upgrading from a version that contains the regression to 1.7.0, 1.6.2 or 1.5.5. If you are using Gloo, Contour, Skipper, or kube-ingress-aws-controller, you shouldn't have any issues. If you use the default "class" (e.g., istio for Istio) for Traefik, Istio, Ambassador, or ingress-nginx, then these should also continue to work without issue.

If you are using Traefik, Istio, Ambassador, or ingress-nginx and you are using a non-default value for the class (e.g., istio-internal), or if you experience any issues with your HTTP-01 challenges please read the notes on Ingress v1 compatibility.

Upgrading with Server Side Apply

As part of the work to remove deprecated APIs cert-manager CustomResourceDefinitions no longer require a conversion webhook. The related change in cert-manager CustomResourceDefinition specs results in invalid CustomResourceDefinition configurations for users who are upgrading to cert-manager 1.7 using kubectl apply --server-side=true -f <manifests>. This can be solved either by performing the upgrade with client side apply or by manually patching the managed fields of cert-manager CustomResourceDefinitions:

crds=("certificaterequests.cert-manager.io" "certificates.cert-manager.io" "challenges.acme.cert-manager.io" "clusterissuers.cert-manager.io" "issuers.cert-manager.io" "orders.acme.cert-manager.io")

for crd in "${crds[@&#8203;]}"; do
  manager_index="$(kubectl get crd "${crd}" --show-managed-fields --output json | jq -r '.metadata.managedFields | map(.manager == "cainjector") | index(true)')"
  kubectl patch crd "${crd}" --type=json -p="[{\"op\": \"remove\", \"path\": \"/metadata/managedFields/${manager_index}\"}]"
done

(Thanks to @​stevehipwell for the above patch commands!)

See the original GitHub issue cert-manager#&#8203;4831

Major Themes

Removal of Deprecated APIs

In 1.7 the cert-manager API versions v1alpha2, v1alpha3, and v1beta1, that were deprecated in 1.4, have been removed from the custom resource definitions (CRDs). As a result, you will notice that the YAML manifest files are much smaller.

In this release we have added a new sub-command to the cert-manager CLI (cmctl upgrade migrate-api-version), which you SHOULD run BEFORE upgrading cert-manager to 1.7. Please read [Removing Deprecated API Resources] for full instructions.

Additional Certificate Output Formats

additionalOutputFormats is a field on the Certificate spec that allows specifying additional supplementary formats of issued certificates and their private key. There are currently two supported additional output formats: CombinedPEM (the PEM-encoded private key followed by the certificate chain) and DER (the DER-encoded private key only). Any combination of output formats can be requested for the same certificate. Read Additional Certificate Output Formats for more details and thanks to @​seuf for getting this across the line!

Server-Side Apply

This is the first version of cert-manager which relies on Server-Side Apply. We use it to properly manage the annotations and labels on TLS secrets. For this reason cert-manager 1.7 requires at least Kubernetes 1.18 (see Supported Releases for further compatibility details).

Configuration Files

In this release we introduce a new configuration file for the cert-manager-webhook. Instead of configuring the webhook using command line flags, you can now modify the webhook Deployment to mount a ConfigMap containing a configuration file. Read the WebhookConfiguration Schema for more information.

In future releases we will introduce configuration files for the other cert-manager components: the controller and the cainjector.

Developing cert-manager Without Bazel

In a future release, we'll remove the use of bazel for building and testing cert-manager, with the aim of making it as easy as possible for anyone to contribute and to get involved with the cert-manager project.

The work is ongoing, but for now we've ensured that cert-manager 1.7 can be built with go build, and that all unit tests can be run with go test ./cmd/... ./internal/... ./pkg/....

Community

Thanks again to all open-source contributors with commits in this release, including:

Thanks as usual to @​coderanger for helping people out on the [#cert-manager Slack channel][#cert-manager Slack channel]; it's a huge help and much appreciated.

In addition, the following cert-manager maintainers were involved in this release:

Changelog since v1.6.0

Feature
  • Add --acme-http01-solver-nameservers flag to enable custom nameservers usage for ACME HTT01 challenges propagation checks. (#​4287, @​Adphi)
  • Add cmctl upgrade migrate-api-version to ensure all CRD resources are stored at 'v1' prior to upgrading to v1.7 onwards (#​4711, @​munnerz)
  • Add goimports verification step for CI (#​4710, @​SgtCoDFish)
  • Add support for loading webhook flags/options from a WebhookConfiguration file on disk (#​4546, @​munnerz)
  • Added additionalOutputFormats parameter to allow DER (binary) and CombinedPEM (key + cert bundle) formats. (#​4598, @​seuf)
  • Added a makefile based build workflow which doesn't depend on bazel (#​4554, @​SgtCoDFish)
  • Added a new Helm chart parameter prometheus.servicemonitor.honorLabels, which sets the honor_labels field of the Prometheus scrape config. (#​4608, @​thirdeyenick)
  • Breaking change: pprof now runs by default on localhost:6060 for webhook and controller, but only if explicitly enabled. Pprof can now be enabled also for cainjector. All three components have --enable-profiling, --profiler-address CLI flags to configure profiling. Thanks to @​bitscuit for help with this! (#​4550, @​irbekrm)
  • Certificate Secrets are now managed by the APPLY API call, rather than UPDATE/CREATE. The issuing controller actively reconciles Certificate SecretTemplate's against corresponding Secrets, garbage collecting and correcting key/value changes. (#​4638, @​JoshVanL)
Bug or Regression
  • Ensures 1 hour backoff between errored calls for new ACME Orders. (#​4616, @​irbekrm)
  • Fix unexpected exit when multiple DNS providers are passed to RunWebhookServer (#​4702, @​devholic)
  • Fixed a bug in the way the Helm chart handles service annotations on the controller and webhook services. (#​4329, @​jwenz723)
  • Fixed a bug that can cause cmctl version to erroneously display the wrong webhook pod versions when older failed pods are present. (#​4615, @​johnwchadwick)
  • Fixes a bug where a previous failed CertificateRequest was picked up during the next issuance. Thanks to @​MattiasGees for raising the issue and help with debugging! (#​4688, @​irbekrm)
  • Fixes an issue in cmctl that prevented displaying the Order resource with cert-manager 1.6 when running cmctl status certificate. (#​4569, @​maelvls)
  • Improve checksum validation in makefile based tool installation (#​4680, @​SgtCoDFish)
  • The HTTP-01 ACME solver now uses the kubernetes.io/ingress.class annotation instead of the spec.ingressClassName in created Ingress resources. (#​4762, @​jakexks)
  • The cmctl experimental install command now uses the cert-manager namespace. This fixes a bug which was introduced in release 1.6 that caused cert-manager to be installed in the default namespace. (#​4763, @​wallrj)
  • Update to latest version of keystore-go to address a backwards-incompatible change introduced in v1.6.0 (#​4563, @​SgtCoDFish)
Other (Cleanup or Flake)
  • Adds clock_time_seconds_gauge metric which returns the current clock time, based on seconds since 1970/01/01 UTC (#​4640, @​JoshVanL)
  • Adds an automated script for cert-manager developers to update versions of kind used for development and testing. (#​4574, @​SgtCoDFish)
  • Breaking change: removes the deprecated dns01-self-check-nameservers flag. Use --dns01-recursive-nameservers instead. (#​4551, @​irbekrm)
  • Bump kind image versions (#​4593, @​SgtCoDFish)
  • Clean up: Remove v1beta1 form the webhook's admissionReviewVersions as cert-manager no longer supports v1.16 (#​4639, @​JoshVanL)
  • Cleanup: Pipe feature gate flag to the e2e binary. Test against shared Feature Gate map for feature enabled and whether they should be tested against. (#​4703, @​JoshVanL)
  • Ensures that in cases where an attempt to finalize an already finalized order is made, the originally issued certificate is used (instead of erroring and creating a new ACME order) (#​4697, @​irbekrm)
  • No longer log an error when a Certificate is deleted during normal operation. (#​4637, @​JoshVanL)
  • Removed deprecated API versions from the cert-manager CRDs (#​4635, @​wallrj)
  • Update distroless base images for cert-manager (#​4706, @​SgtCoDFish)
  • Upgrade Kubernetes dependencies to v0.23.1 (#​4675, @​munnerz)

v1.6.3

Compare Source

v1.6.3 Release Notes

1.6.3 is a minor release rebuilding cert-manager 1.6 using the latest version of Go. This eliminates a few security vulnerabilities which have accumulated in Go since the last release.

We don't believe any of those vulnerabilities were practically exploitable or relevant to cert-manager, but we decided to rebuild to keep up to date anyway.

Changelog since cert-manager 1.6.2

Bug or Regression
  • Bumps the version of Go used to build the cert-manager binaries to 1.17.8, to fix a slew of CVEs (none of which were likely to be exploited) (#​4975, @​vhosakot)
  • Fixes an expired hardcoded certificate which broke unit tests (#​4977, @​SgtCoDFish @​jakexks)

v1.6.2

Compare Source

In 1.6.2, we have reverted a change present in 1.6.0 and 1.6.1 that caused a regression in the ACME Issuer. In 1.6.0 and 1.6.1, the Ingress created by cert-manager while solving an HTTP-01 challenge contained the kubernetes.io/ingress.class annotation:

apiVersion: networking.k8s.io/v1beta1
kind: Ingress
metadata:
  annotations:
    kubernetes.io/ingress.class: istio # The `class` present on the Issuer.

After 1.5, the Ingress does not contain the annotation anymore. Instead, cert-manager uses the ingressClassName field:

apiVersion: networking.k8s.io/v1
kind: Ingress
spec:
  ingressClassName: istio # 🔥 Breaking change!

This broke many users that either don't use an Ingress controller that supports the field (such as ingress-gce and Azure AGIC), as well as people who did not need to create an IngressClass previously (such as with Istio and Traefik).

The regression is present in cert-manager 1.5.4, 1.6.0, and 1.6.1. It is only present on Kubernetes 1.19+ and only appears when using an Issuer or ClusterIssuer with an ACME HTTP-01 solver configured.

In 1.6.2, we have restored the original behavior which is to use the annotation. This patch is also available in 1.5.5 and in 1.7.0.

Most people won't have any trouble upgrading from 1.6.0 or 1.6.1 to 1.6.2. If you are using Gloo, Contour, Skipper, or kube-ingress-aws-controller, you shouldn't have any issues. If you use the default "class" (e.g., istio for Istio) for Traefik, Istio, Ambassador, or ingress-nginx, then these should also continue to work without issue.

If you are using Traefik, Istio, Ambassador, or ingress-nginx and you are using a non-default value for the class (e.g., istio-internal), or if you experience any issues with your HTTP-01 challenges please read the notes on Ingress v1 compatibility.

Changelog since v1.6.1

Bug or Regression
  • The HTTP-01 ACME solver now uses the kubernetes.io/ingress.class annotation instead of the spec.ingressClassName in created Ingress resources. (#​4785, @​jetstack-bot)
Other (Cleanup or Flake)
  • cert-manager now does one call to the ACME API instead of two when an Order fails. This fix is part of the effort towards mitigating the high load that cert-manager deployments have on the Let's Encrypt API (#​4619, @​irbekrm)
  • Bump base images to latest versions (#​4707, @​SgtCoDFish)

Dependencies

Added

Nothing has changed.

Changed

Nothing has changed.

Removed

Nothing has changed.

v1.6.1

Compare Source

Changelog since v1.6.0

Changes by Kind

Bug or Regression
  • Fixes an issue in cmctl that prevented displaying the Order resource with cert-manager 1.6 when running cmctl status certificate. (#​4572, @​maelvls)
  • Update to latest version of keystore-go to address a backwards incompatible change introduced in v1.6.0 (#​4564, @​SgtCoDFish)

Dependencies

Added

Nothing has changed.

Changed
Removed

Nothing has changed.

v1.6.0

Compare Source

Changelog since v1.5.0

Breaking Changes (You MUST read this before you upgrade!)

⚠️ Following their deprecation in version 1.4, the cert-manager APIVersions v1alpha2, v1alpha3, and v1beta1 are no longer served.

This means if your deployment manifests contain any of these API versions, you will not be able to deploy them after upgrading. Our new cmctl utility or old kubectl cert-manager plugin can convert old manifests to v1 for you.

⚠️ JKS Keystores now have a minimum password length of 6 characters, as an unintended side effect of upgrading keystore-go from v2 to v4. This was fixed in cert-manager v1.6.1

Changes by Kind

Feature
Bug or Regression
  • FIX: Prevent Vault Client from panicing when request to Vault health endpoint fails. (#​4456, @​JoshVanL)
  • Fix CRDs which were accidentally changed in cert-manager v1.5.0 (#​4353, @​SgtCoDFish)
  • Fix regression in Ingress PathType introduced in v1.5.0 (#​4373, @​jakexks)
  • Fixed the HTTP-01 solver creating ClusterIP instead of NodePort services by default. (#​4393, @​jakexks)
  • Fixes renewal time issue for certs with skewed duration period. (#​4399, @​irbekrm)
  • Pod Security Policy for startup api check job (#​4364, @​ndegory)
  • The startupapicheck post-install hook in the Helm chart now deletes any post-install hook resources left after a previous failed install allowing helm install to be re-run after a previous failure. (#​4433, @​wallrj)
  • The defaults for leader election parameters are now consistent across cert-manager and cainjector. (#​4359, @​johanfleury)
  • Use GetAuthorization instead of GetChallenge when querying the current state of an ACME challenge. (#​4430, @​JoshVanL)
Other (Cleanup or Flake)

v1.5.5

[Compare Source](https://togithub.com/jetstack/cert-

Configuration

📅 Schedule: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined).

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

  • If you want to rebase/retry this PR, click this checkbox.

This PR has been generated by Mend Renovate. View repository job log here.

@renovate renovate bot force-pushed the renovate/cert-manager-1.x branch from f521a28 to 24d68cd Compare June 14, 2022 18:55
@renovate renovate bot changed the title Update Helm release cert-manager to v1.8.0 Update Helm release cert-manager to v1.8.1 Jun 14, 2022
@renovate renovate bot force-pushed the renovate/cert-manager-1.x branch from 24d68cd to 9365d4d Compare June 23, 2022 20:25
@renovate renovate bot changed the title Update Helm release cert-manager to v1.8.1 Update Helm release cert-manager to v1.8.2 Jun 23, 2022
@renovate renovate bot changed the title Update Helm release cert-manager to v1.8.2 Update Helm release cert-manager to v1.8.2 - autoclosed Jun 24, 2022
@renovate renovate bot closed this Jun 24, 2022
@renovate renovate bot deleted the renovate/cert-manager-1.x branch June 24, 2022 13:26
@renovate renovate bot changed the title Update Helm release cert-manager to v1.8.2 - autoclosed Update Helm release cert-manager to v1.8.2 Jun 27, 2022
@renovate renovate bot restored the renovate/cert-manager-1.x branch June 27, 2022 06:26
@renovate renovate bot reopened this Jun 27, 2022
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

None yet
0 participants