Key details

  1. Released October 26, 2026 as first stable Stalwart server version.
  2. New on-disk schema requires migration; old stores cannot be opened by 1.0.
  3. v0.16.27 is the prerequisite for stalwart-migrate; older versions use documented alternative migration path.
  4. Cluster coordination now requires Redis; single-node has built-in local store.
  5. New document-attachment full-text indexing and SIMD/SWAR parsing.
  6. Vendor reports 35/38 benchmark wins against Postfix 3.11/Dovecot 2.4 with scripts and raw results published.
  7. Post-quantum DKIM2, OAuth client metadata and finer scopes added.

What builders should take away

  1. Back up mail and test an incremental migration from v0.16.27 before production cutover.
  2. Do not replace a pre-1.0 binary without preparing a new v1.0 data store.
  3. For clustered installations, plan Redis coordinator availability and failover.
  4. Test OAuth clients, IMAP UIDs, mail delivery, search rebuild and rollback after migration.
  5. Treat vendor performance numbers as a reproducible claim, not an independent guarantee.
  6. Check Community versus Enterprise feature boundaries before choosing a deployment.

What changed

Stalwart released v1.0 on October 26, 2026, its first stable mail/collaboration server release after five years. The Rust-based server supports SMTP, IMAP, POP3, JMAP, CalDAV, CardDAV and WebDAV. Version 1.0 finalises the on-disk database layout, substantially rewrites MIME parsing and storage/search paths, adds full-text search inside document attachments and expands authentication controls. Its maintainers publish benchmark scripts and report winning 35 of 38 comparisons against a configured Postfix 3.11 and Dovecot 2.4 stack, but those results remain vendor-run. Crucially, v1.0 cannot directly open pre-1.0 data stores: v0.16 operators must first reach 0.16.27 and run stalwart-migrate; older versions can use a migration proxy and Vandelay. Clusters now require Redis coordination.

Why it matters

For a small team operating its own mail, the difference between routine binary updates and mandatory database conversion determines downtime, rollback planning and data-integrity risk. The stabilised schema could reduce future upgrade complexity, but getting to 1.0 requires a tested migration. Performance claims are unusually well documented with reproducible benchmark configurations, yet remain maintainer measurements rather than independent proof that Stalwart is universally faster. Teams evaluating a move from Postfix/Dovecot also need to consider operational maturity, mail deliverability, security, client compatibility and support boundaries.

Stable storage is the practical milestone

The 1.0 database schema is intentionally finalised so later releases can upgrade in place rather than repeating disruptive migrations. This release itself is a breaking storage transition: the new server will not open an old store. The official migration tool reads the old store, copies data to a new one, supports incremental reruns and preserves IMAP UIDs and change history. The final cutover still needs planning; OAuth users must sign in again and search indexes rebuild.

How existing installations move

For v0.16 installations, update to 0.16.27, back up and test, then use stalwart-migrate with a new empty v1.0 store. For v0.15 or earlier and gradual moves, the documented migration proxy and Vandelay can route accounts between old and new servers. Clusters must now use a Redis-family in-memory coordinator; single-node installations can use a built-in local store. Configuration keys were also renamed or restructured.

Performance and search changes are substantial but vendor-measured

The maintainers rewrote mail-parser using SIMD/SWAR, tuned RocksDB, PostgreSQL, MySQL, FoundationDB and SQLite backends, and redesigned mailbox caches. A new full-text index supports phrase searches and text extraction from PDF, Office, OpenDocument, EPUB and RTF attachments. On matched Azure servers, Stalwart says it won 35 of 38 head-to-head Postfix/Dovecot tests. Benchmark scripts and raw data are published, but independent reproduction and comparisons across different production configurations are still needed.

Security, protocols and integration caveats

The release includes JMAP object metadata, IMAP METADATA, more OAuth/OIDC controls and post-quantum DKIM2 support. Early community upgrade reports noted a Bulwark Native OAuth client ID issue, with a webmail fix published and native update expected. Administrators should test every mail client, authentication flow and relay before switching production. Enterprise-only AI spam classification should not be presented as a Community-edition feature.

What to watch next

  • Independent performance replication against comparable Postfix/Dovecot configurations.
  • Migration and OAuth client compatibility reports from early adopters.
  • Whether future 1.x upgrades avoid data migrations as promised.

Still unclear

  • The 35/38 performance comparison is maintainer-run; no independent replication was established.
  • New stable-schema promises cannot yet be validated against future releases.
  • Early OAuth-client compatibility fixes may still be propagating across native clients.

Sources

Direct reading behind this dossier.

3 sources

Discussion

Discussion is reader-contributed. Comments are not part of the BTN dossier or its editorial evidence.

0 visible comments

Join the discussion

Keep comments useful and relevant. Reader contributions may be moderated and are not BTN editorial evidence.

Sign in to comment