Document URL: https://docs.openshift.com/container-platform/3.7/install_config/redeploying_certificates.html#redeploying-master-certificates Section Number and Name: "Installation and Configuration" > "Redeploying Certificates" > "Redeploying Master Certificates Only" Describe the issue: When following "Redeploying Master Certificates Only" documentation [1] there is no indication service-signer.crt and service-signer.key (used for "Service Serving Certificates" [2]) will be purged and regenerated. This is the CA cert responsible for signing any services with annotation "service.alpha.openshift.io/serving-cert-secret-name." It's not obvious this will happen without examining function certificates_to_synchronize [3]. Following execution of redeploy-master-certificates.yml the certificates previously signed by the original service-signer.crt are invalid. [1]: https://docs.openshift.com/container-platform/3.7/install_config/redeploying_certificates.html#redeploying-master-certificates [2]: https://docs.openshift.com/container-platform/3.7/dev_guide/secrets.html#service-serving-certificate-secrets [3]: https://github.com/openshift/openshift-ansible/blob/master/roles/lib_utils/filter_plugins/openshift_master.py#L495-L496 Suggestions for improvement: Until redeploy of service-signer.crt and service-signer.key (used for "Service Serving Certificates" [2]) is split-off from redeploy-master-certificates.yml there should be a warning or informational disclaimer noting "Service Signing Certificates" will need to be regenerated after playbook execution (by deleting existing secrets containing service serving certificates or removing/re-adding annotations to appropriate services). Additional information: Due to recreation of service-signer.crt & service-signer.key implicit trusts required for proper routing will break. Router pods will stop trusting old certificates (signed by the original service-signer.crt) still referenced in running pod secrets and as a result HAProxy may mark the backends as down due to "Layer6 invalid response, info: SSL handshake failure"
@j.r. -- PTAL https://github.com/openshift/openshift-docs/pull/8373 It doesn't appear from the BZ or the https://access.redhat.com/support/cases/02041417 that we need to add the Additional Information from the description?
(In reply to Michael Burke from comment #2) > @j.r. -- PTAL > > https://github.com/openshift/openshift-docs/pull/8373 > > It doesn't appear from the BZ or the > https://access.redhat.com/support/cases/02041417 that we need to add the > Additional Information from the description? Correct, Michael. LGTM - the [Important] note is perfect. Thank you!
Add comments in the PR
Changes are live: https://docs.openshift.com/container-platform/3.9/install_config/redeploying_certificates.html#redeploying-master-certificates