Skip to content
INSIGHTS & GUIDES · NETCLOUD24

Making a certificate revocation reach the relying application

Carry one approved certificate revocation from issuer and serial verification through reason selection, CRL publication, endpoint delivery and application validation, while preserving caching, replacement and retirement limits.

Read the article ↓
Editorial illustration: How to Revoke a Certificate and Publish a New CRL

Platform: Windows Server 2019, 2022 and 2025, with the version and hosting restrictions described below.

Identify the credential and the security decision

Revocation withdraws a certificate before its normal expiry. The first task is identifying the exact issuer and serial number and establishing why it must no longer be trusted. A similar subject name or the newest certificate in a store is not enough. Record the service, requester, template, request identifier and incident or change authorisation before altering the CA database.

For managed Netcloud24 environments, the team handles the CA operation and distribution checks within the agreed PKI responsibility. Customers should report the affected service or suspected exposure promptly rather than attempt mass certificate deletion. This workflow follows a single authorised revocation through publication and client verification, recognising that the decision is not instantly visible everywhere.

Distinguish revocation, replacement and deletion

Removing a certificate from IIS stops that binding using it, but does not tell another relying system it is revoked. Deleting a local certificate is also not a CA revocation record. Issuing a replacement does not automatically invalidate the predecessor. These are separate lifecycle actions that may all be required in an incident.

Select the reason accurately. Key compromise and CA compromise have different incident implications from superseded or cessation of operation. Certificate Hold is a deliberate temporary status and is the reversible reason; it should not be used to disguise an irreversible compromise decision or as a routine retry mechanism. Most other revocations cannot simply be undone.

For a compromised CA key, follow the hierarchy incident plan as well as the certificate operation. A normal leaf-certificate replacement is not sufficient to repair trust in a compromised issuer. Preserve evidence and coordinate affected services rather than changing arbitrary certificates with similar names.

Know how clients learn about the decision

The CA records revocation and publishes signed data. A base CRL contains the relevant revocation list; a delta, where configured and supported, supplements its associated base. A publication path is where the CA writes files, while a CDP is a retrieval location embedded in certificates. A local file-generation success does not prove the retrieval service received the new version.

Clients and applications can cache revocation information and follow different validation policies. A revoked certificate may remain accepted until newer information is retrieved, or an unreachable CRL can cause a validation failure according to application policy. Do not claim immediate universal blocking from a database change alone.

Keep old CDP URLs operational while unexpired certificates still reference them. Changing the CA’s current extension settings affects later certificates; it does not rewrite existing ones. This is particularly important during server migrations and publication-host changes.

Revoke the exact serial on the issuing CA

Use a certificate-manager identity with the appropriate Issue and Manage Certificates permission. CA configuration changes and publication scheduling belong to the corresponding CA-administration role. In Certification Authority, locate the issued certificate, verify its identity and select the approved revoke action and reason.

A controlled command alternative is shown for a confirmed key-compromise decision. The serial placeholder must be replaced by the exact issuer-scoped value; code 1 is not a generic cleanup reason. Review the CA configuration name carefully when more than one issuer exists.

# Irreversible for this reason: use only for the approved incident.
certutil -config 'ISSUE01.corp.example\Example Issuing CA' `
  -revoke 'REPLACE_WITH_VERIFIED_SERIAL_NUMBER' 1

Record the effective time, reason and result. Do not bulk-revoke every certificate in a console view to fix one service. If the affected private key is compromised, ensure the replacement uses a new protected key rather than merely renewing the same exposed material.

Publish and deliver current revocation files

Publish the relevant CRL data through the approved CA workflow. Then verify the local output, the transfer job and every required retrieval endpoint separately. For an offline root, that includes the controlled media and publication ceremony rather than assuming an online network copy.

certutil -config 'ISSUE01.corp.example\Example Issuing CA' -CRL
# Inspect an approved local copy of the published public CRL:
certutil -dump 'C:\PKI-Evidence\Example-Issuing-CA.crl'

Inspect issuer, signature, CRL number, this-update and next-update and confirm the serial appears with the intended reason. If deltas are advertised, verify the required base and delta versions both arrive. Protect the publication writer while keeping client retrieval available without inappropriate authentication prerequisites.

A stale web copy can remain after the CA generated a fresh CRL. Check actual file bytes and serving behaviour, not just a green replication-job icon. Consider caching layers and web MIME handling when clients cannot retrieve a file that appears present on disk.

Validate through a representative relying path

Use a public copy of the affected certificate and a representative client network. Inspect its exact CDP URLs and retrieve the current information through the real path. A laboratory client with a different resolver or trust store can give a misleading result.

certutil -dump 'C:\PKI-Evidence\AffectedCertificate.cer'
certutil -urlfetch -verify 'C:\PKI-Evidence\AffectedCertificate.cer'

The verification should report the expected revoked state once it consumes the appropriate current data. Then test the actual application’s validation behaviour using an authorised safe method. Do not clear every certificate cache across production merely to make the change appear instantaneous; understand the application’s caching and incident containment options.

For urgent compromise, additional containment may be necessary while revocation propagates, such as removing the exposed binding or disabling the affected access path. Coordinate that with the service owner rather than relying solely on an assumed client refresh time.

Diagnose stale trust at the right stage

If the certificate is still accepted, confirm the issuer and serial, CA record, CRL publication, transfer, endpoint freshness and client cache in that order. If validation fails because retrieval is unavailable, check DNS, network scope, authentication requirements and the exact referenced URL. A new working URL does not fix older certificates that still point elsewhere.

A client fetching the base but failing on a delta needs a consistent configured pair. Do not suppress revocation checks globally to conceal a distribution fault. If the certificate was placed on hold and is legitimately reinstated, publish newer revocation data and validate again; cached information can continue reporting the old state until refreshed.

Retain the lifecycle evidence and publication obligations

Keep approval, serial, reason, publication versions and verification evidence with the incident. If the wrong certificate was permanently revoked, issue a replacement through the correct process; do not promise an unsupported reversal. Preserve the old endpoint until no relevant valid credential depends on it.

Monitor CRL freshness and expiry continuously, including root and issuing tiers. Renewal and retirement planning must include the longest remaining certificate references, not merely whether the CA server is still used for new requests. A managed revocation operation is complete when the decision is recorded, distributed and observed by the intended relying service, with remaining cache or containment limits clearly understood.

Further reading


Was this answer helpful?

« Back
Contact

Write to us

Do you have questions about our services or need a quote? Write to us and we will respond as soon as possible.

Contact form

Fields marked * are required.