[ Sending-service guides ]

Authenticate email sent through Zimbra: configuration and verification

Configure the documented Zimbra sending route, preserve other mail services, and verify the delivered identities before accepting the change.

By the EmailAuth research team · 5 min read

[ Key points ]

  • Configure the documented Zimbra sending route, preserve other mail services, and verify the delivered identities before accepting the change.
  • A local key-pair check cannot prove the final outbound message.
On this page
  1. Scope and prerequisites
  2. Configure the documented route
  3. Product-specific risks
  4. Prove the route at the receiving end
  5. Diagnose failures without widening access
  6. Handover and recovery
  7. Confirm the provider boundary
  8. Work through a fictional diagnostic case
  9. Separate a passing message from enforcement readiness

Scope and prerequisites

This guide covers an administered Zimbra Collaboration MTA, not an unrelated hosted mailbox branded Zimbra. Confirm the installed version, outbound relays and who controls the MTA and DNS before changing signing. The first-party wiki documents domain-level DKIM, including alias domains.

Configure the documented route

  • Step 1. On the MTA, inspect the domain's existing configuration before creating keys. Protect diagnostic output: zmdkimkeyutil -q can disclose the private key.
  • Step 2. For a domain without DKIM, the documented generation command is /opt/zimbra/libexec/zmdkimkeyutil -a -d example.com, run in the appropriate Zimbra administration context. Replace the example domain with the approved domain.
  • Step 3. Publish only the generated public DNS record at the generated selector under _domainkey. Verify the TXT answer at authoritative DNS.
  • Step 4. Use the version-appropriate opendkim-testkey location to check the key pair, then test an outbound message.
  • Step 5. For rotation, the documented update option is -u. Retain the previous public record long enough for previously signed messages to verify. Do not treat -r, which removes signing data, as a rotation command.

Product-specific risks

Prepare a maintenance window because key generation and DNS publication do not become visible everywhere simultaneously. If an outbound relay rewrites a signed body, a locally valid key can still yield a failed received signature. Place the test after the entire outbound route, including disclaimers and security gateways. For a hosted deployment without MTA access, give the hosting operator the test requirements rather than attempting shell commands on a workstation. Never attach key-query output to a support ticket without removing secrets.

Prove the route at the receiving end

Use an approved external mailbox and save the complete raw message with its timestamp and message identifier. No live tenant test is claimed in this guide. The following is an acceptance worksheet to complete during implementation, not a record of observed results.

  • Evidence field: Visible From; What to record: Actual address and domain; Acceptance question: Is this the intended sender identity?
  • Evidence field: Envelope; What to record: Return-Path and receiver-reported SPF domain; Acceptance question: Which domain did SPF actually authenticate?
  • Evidence field: SPF; What to record: Trusted receiver result and evaluated identity; Acceptance question: Did it pass, and does that identity align?
  • Evidence field: DKIM; What to record: Every d= domain, s= selector and receiver result; Acceptance question: Does at least one passing signature align?
  • Evidence field: DMARC; What to record: Receiver result and applicable policy; Acceptance question: Does an aligned passing path exist?
  • Evidence field: Route; What to record: Product, account and message type; Acceptance question: Did the test use the production path?

DMARC needs at least one aligned, passing SPF or DKIM path. A passing signature from an unrelated service domain is not enough. Strict alignment requires exact domain equality; relaxed alignment uses the applicable organizational-domain relationship. Check the domain's actual alignment settings before declaring success. A subdomain is not an automatic strict match. SPF uses the SMTP identity, not the display name or Reply-To. Consult the alignment glossary when recording the decision.

Use Authentication-Results from the receiving system you trust, not a header supplied by an arbitrary sender. Capture all signatures rather than stopping at the first failure. For forwarded mail, compare a direct test with a forwarded copy: forwarding may change SPF evaluation or alter signed content. Do not diagnose those effects by adding the forwarder's address to your provider authorization.

Diagnose failures without widening access

Compare authoritative DNS with the exact configured record names first, then check recursive visibility and service validation. A wrong name, truncated key or conflicting CNAME is not fixed by waiting for caches. Keep only one SPF policy at an owner and check the ten DNS-lookup-term limit across evaluated includes and redirects before merging another sender. DKIM selectors belong to individual signing configurations; do not overwrite another application's record to force a validation result.

If authentication passes but delivery is poor, inspect the recipient response, reputation, content and sending permissions separately. DMARC pass does not promise acceptance or inbox placement. If neither path aligns, stop expansion and correct the actual sender configuration. Do not weaken the whole domain's DMARC policy to hide a single product failure.

Handover and recovery

Save the approved DNS diff, public lookup results, service status and redacted received-message worksheet with a named owner. Keep credentials and private keys out of tickets. On failure, pause the affected sending job or use a previously tested route under the service owner's control. Restore only the changed records that belong to this integration; keep public keys needed by in-flight messages. Retest after rollback, key rotation, domain migration or a change in outbound relay. Close the change only when the product owner and mail administrator can explain the observed identity chain.

Confirm the provider boundary

The wiki identifies different test-tool locations for ZCS 8.6 and earlier versus 8.7 and later, and says 8.7.x generates 2048-bit keys. Confirm the installed version rather than copying an old sample key. For a long TXT value, use multiple character strings within one resource record, not competing public-key records. The wiki warns about 8-bit conversion and mailing-list/PGP interactions; test those applicable routes without casually changing encoding or restarting services. Revocation with an empty p= invalidates that key and is not ordinary overlap rotation.

Work through a fictional diagnostic case

Fictional written case only: ZM-01 assumes local key testing succeeds for shop.example, but a downstream gateway adds a footer and the received message from help@shop.example reports a body-hash failure for d=shop.example. SPF and other signatures are unknown. The MTA and gateway owners must establish the final signing order and test after all modifications; a local test does not contradict the received failure. No full DMARC verdict or repaired delivery is asserted. No tenant, DNS, server command or message was exercised for this case.

Separate a passing message from enforcement readiness

RFC 9989 Section 7.4 requires valid DKIM for a domain publishing p=reject and cautions about indirect-mail interoperability. A single aligned SPF result does not by itself justify a reject-policy rollout. Review all legitimate routes and policy ownership separately from this provider setup.

[ Takeaway ]

A local key-pair check cannot prove the final outbound message.

[ By role ]

What this means for you

Choose your role for practical next steps.

IT / Mail Admin

Verify identities, not green badges

For Zimbra, keep product activation, DNS visibility and receiver results as separate evidence.

Next steps

  1. Capture current authoritative records before editing.
  2. Test each distinct sender and delivery route.
  3. Retain the public evidence and a narrow rollback plan.

CISO / SecOps

Limit delegated sending authority

The Zimbra integration should have a documented domain scope and accountable owner.

Next steps

  1. Review selector and envelope ownership before delegation.
  2. Exclude credentials and private keys from handover material.
  3. Require revalidation after channel or domain changes.

Founder / Non-technical

Protect the message that matters

Assign business ownership to the Zimbra route rather than treating DNS validation as project completion.

Next steps

  1. Name the critical message type and its sender.
  2. Approve a small representative test before broader sending.
  3. Keep an operational fallback and owner for failures.

[ FAQs ]

Frequently asked questions

Should a hosted-mailbox user run server commands?

No. These steps require authority over the Zimbra MTA. A hosted deployment needs its operator to perform the server-side change.

What should I never put in the evidence ticket?

Private keys, credentials or unredacted diagnostic output that contains them. Public DNS answers and receiver headers are normally enough for authentication diagnosis.

Can a relay break an otherwise valid signature?

Yes. Changes to signed content after signing can make DKIM fail. Test a message after its complete outbound processing path.

Does successful key lookup prove production success?

No. It proves DNS availability at that query point. A representative received message must also carry a valid, aligned signature.

What proves Zimbra mail passes DMARC?

A representative received message must have at least one passing SPF or DKIM identifier aligned with its visible From domain under the applicable alignment mode. Service verification alone is insufficient.

Can Zimbra authentication guarantee the inbox?

No. The recipient retains its own acceptance and filtering policy. Keep authentication results separate from placement and reputation evidence.

What should I save for a Zimbra handover?

Save the product route, approved DNS diff, verification status and redacted receiver evidence. Include the timestamp and sender identity, but exclude credentials and private keys.

When should the Zimbra route be retested?

Retest after a domain or selector change, relay migration, sending-channel change or rollback. Cover each distinct sender domain and representative message type.

[ Sources ]

Sources

  1. Configuring for DKIM Signing - Zimbra :: Tech Center

    Sections The zmdkimkeyutil utility · Updating DNS · §2048-bit signatures starting ZCS 8.7.x · DKIM and outbound email · Revoking a DKIM key in DNS

  2. dmarc

    Sections §4.4 · §5.3.5 · §7.4

  3. spf

    Sections §4.5 · §4.6.4

  4. dkim

    Sections §3.5 · §3.6.1 · §6.1

[ Glossary ]

Glossary

Alignment
The required relationship between the visible From domain and an authenticated SPF or DKIM domain.
Selector
The DKIM signature identifier that selects a DNS public key beneath the signing domain.
Return-Path
The delivered representation of the SMTP reverse-path, useful alongside receiver results when identifying the SPF domain.

Enterprise provenance

Why top financial institutions and governments retired their email security dashboards.

Traditional email authentication platforms sell you a dashboard and ask your engineers to do the labour. In high-stakes environments, manual oversight means delayed enforcement, configuration drift and unmonitored shadow domains.

EmailAuth is the agentic email provenance platform. Our agents enforce p=reject, keep third-party senders aligned and move your domains to full RFC 9989 compliance, the new DMARC standard,, with human-in-the-loop auditability.

Governments Large enterprises Tier-1 financial institutions

SaaS · Dedicated tenancy · Air-gapped on-prem · BYOK encryption

Knowledge Base