Mailcow 2026-11 patches an RCE and adds mailbox groups
On November 5, 2026, Mailcow shipped its 2026-11 update: it fixes a remote code execution vulnerability exploitable by authenticated admins and domain admins, and adds mailbox groups, directory sync, and multiple API keys. The update is urgent, and exposed instances should restrict admin access in the meantime.
November 5, 2026. Mailcow ships its 2026-11 update, dubbed “Moovember”. November 5, 2026. It fixes a remote code execution (RCE) vulnerability exploitable by authenticated admins and domain admins. November 5, 2026. The technical details will not be published until next week; until then, the vendor recommends restricting access to admin accounts. The whole thing fits in one sentence: the most widely installed self-hosted mail suite closes a door into the server while adding exactly what it lacked for managing teams.
For an organization running its own mail with Mailcow, this is not just a changelog. It is proof that the control plane of a mail suite — the administration interface — is an attack surface of its own, separate from the Postfix and Dovecot services it orchestrates.
The flaw: an RCE reserved for authenticated admins
The release notes are unusually direct. Mailcow states it is fixing a remote code execution vulnerability “exploitable by authenticated admins and domain admins”, without yet providing a CVE identifier or CVSS score — both will follow “early next week”. Until then, the instruction is simple: limit access to admin and domain admin accounts.
That wording deserves a precise reading. A flaw that requires administrator privileges is not harmless, appearances aside. First, because in a mail suite the admin already holds everything: mailboxes, aliases, quotas, relays, backups. A compromised admin is already a near-total compromise of the mail system. Second, because an RCE turns a high-privilege account into command execution on the host running the Docker containers — a perimeter jump that merely editing mailboxes does not allow.
The fix ships alongside two cross-cutting hardening measures.
- Fail2ban now covers every protocol — IMAP, POP3, SMTP, Sieve, EAS, DAV — rather than only web authentication failures.
- 2FA is now rate-limited, making brute-force of one-time codes materially harder.
The underlying message is clear: Mailcow is hardening the authentication layer at the very moment it patches a flaw that feeds on it.
What the update adds: groups, directory sync, API keys
2026-11 is more than a security release. It adds three building blocks that pull Mailcow closer to a genuine team mailbox platform.
- Mailbox groups. Distribution lists, send-as, shared mailboxes, and group-shared SOGo calendars and address books. This is the missing piece for replacing a minimal Google Workspace or Microsoft 365 on shared use.
- Directory sync. Groups and their members can be synced from Keycloak, LDAP, or Active Directory, with optional offboarding. A departure in the directory can now revoke mail access automatically instead of relying on a manual procedure.
- Multiple API keys. Each key can be read-only or read-write, with its own IP allowlist. Legacy keys from
mailcow.confappear as locked service keys.
That last block is the most structural for a serious self-hoster. A single, all-powerful API key was a latent risk: any integration holding it could break or exfiltrate everything. Multiple keys with scoped permissions finally bring least-privilege to the scripts that drive the platform.
The breaking changes worth a second look
Any update that reworks the administration surface breaks automation. 2026-11 is no exception, and two changes deserve a review before docker compose pull.
- Sync job API. The
add/syncjobendpoint now requires asource_id: the oldhost1,port1,enc1,authmech1fields are no longer accepted or returned. Existing jobs migrate automatically, but any script that creates or reads sync jobs must create a source viaadd/syncjob_sourceand pass itssource_id. - Request body limit. The limit becomes 64 MB for the UI and API, with
NGINX_MAX_BODY_SIZE(default128m) for SOGo, ActiveSync, and DAV. Instances that raised the Postfix attachment size limit must setNGINX_MAX_BODY_SIZEinmailcow.confand rundocker compose up -d.
The point to watch is not the severity of these changes — they are documented — but their invisibility: a request body limit breaks nothing at first restart, then returns 413 errors on heavy uploads weeks later, usually at the worst moment.
What it changes for self-hosted email
Mailcow occupies a singular place in self-hosting: it is one of the few complete mail suites — Postfix, Dovecot, Rspamd, SOGo, ClamAV — bundled into a single docker compose. That position gives it a large install base, but it also concentrates responsibility: a flaw in one component, and an organization’s entire email is exposed.
The RCE in 2026-11 is a reminder of a truth that gets forgotten. Hosting your own mail is not only about configuring SPF, DKIM, and DMARC once and for all. It is also about updating the control plane — the web UI, SOGo, the API — with the same discipline as the mail daemons. A mail suite that is not updated becomes a target, not because Postfix is vulnerable but because the administration layer around it is.
Mailcow’s cadence — monthly, with regular security fixes — is both a strength and a constraint. A strength, because flaws get fixed fast. A constraint, because an instance abandoned for six months can accumulate several exploitable flaws and a pile of breaking changes to absorb at once.
Verdict
Mailcow 2026-11 is first and foremost a security release: the RCE, even restricted to authenticated admins, justifies updating this week on any exposed instance, and restricting admin accounts immediately where patching cannot happen right away.
If you run Mailcow, apply 2026-11 without waiting for the CVE, then re-check your API scripts for the source_id change. If you are weighing a self-hosted team mailbox, the groups and directory sync now make Mailcow a credible alternative to a minimal tenant — provided you accept the monthly update discipline that comes with it. The trap is not the complexity of the suite — it is remarkably contained — but the temptation to deploy it and then forget it, as if email administered itself.