YariloMail 2.4.4: a Submission fix, JMAP submission and Sieve include parity ​

19 October 2026

YariloMail 2.4.4 is a patch release of the 2.4 stable line. Upgrade promptly if clients send mail through the Submission login: in 2.4.3, after AUTH, every command got the previous command's reply, so no conforming client could send.

Security ​

  • With ihave in sieve_extensions, require "ihave" let a user's script use every extension of the build, whatever the list said. Only installations that had added ihave to sieve_extensions were exposed; the default list does not include it. Fixed.

New ​

  • JMAP submission: EmailSubmission and Identity over the Submission relay, VacationResponse as the managed Sieve script ManageSieve also sees, and SieveScript (RFC 9661). Identities come from the login address and the new userdb field identity_aliases.
  • Submission relay can log in as a master with the user as authzid, and verify the relay against a named CA.
  • Sieve: include :global, with the chart mounting global scripts; sieve_global_extensions for extensions only the administrator's scripts may use; ORCPT for vacation.
  • ManageSieve quotas: a script count and a storage limit, with HAVESPACE.

Fixed ​

  • Submission login: replies no longer shift by one command after AUTH.
  • Sieve: include runs at delivery and in imapsieve; vacation follows RFC 5230 more closely, with nine new sieve_vacation_* settings.
  • Helm: imapsieve scripts reach the imap container of the co-located backend.
  • Director and JMAP: a set of fixes; the release notes list them.

Before you upgrade ​

  • general.xclient.trusted_nets now defaults to empty. If your values never set it and you use XCLIENT, name the proxies' networks.
  • Global before/after Sieve scripts are checked against sieve_extensions plus sieve_global_extensions when they load; one that uses more is left out.
  • ihave answers from the whitelist.
  • Config schema 10 → 17: upgrade the chart and the image together.
  • Pull the new image onto the backend nodes before helm upgrade, and keep room for one more backend pod.

Read the release notes before you run helm upgrade.

Get it ​