Dolt v2.2.0 - JWT Authentication Contract Change

Dolt v2.2.0 was published on July 15, 2026. The release draws a new version boundary around a change to JWT token authentication through authentication_dolt_jwt. It contains no newly merged pull requests beyond what shipped in 2.1.11, so compatibility testing matters more than feature evaluation.

The full release notes and downloads are on the GitHub release page.

The project describes v2.2.0 as nearly identical to 2.1.11. The minor version bump exists because token authentication through authentication_dolt_jwt changed in 2.1.11. This release makes that compatibility boundary explicit even though it does not add a fresh set of merged changes.

The notes say the authentication change could affect deployments that mint JWT tokens for Dolt. They also say existing deployments are not expected to be affected. For data platforms, the useful distinction is between a running connection path that already works and automation that creates fresh tokens for new workers, scheduled jobs, or replacement instances.

The notes do not specify which token fields or validation rules changed. They also do not provide a migration recipe. Test with tokens produced by the same issuer and configuration used in production, and include token renewal in that test. The v2.2.0 release page is the source record for the compatibility notice.

There are no novel merged pull requests in v2.2.0. That limits the review surface, but it does not make the release irrelevant. The version number communicates that the authentication behavior already present in 2.1.11 deserves explicit operator attention.

Do not expect a new SQL engine feature, storage format change, or command surface from this tag based on the supplied notes. Teams already running 2.1.11 have likely crossed the authentication boundary. Teams upgrading from an earlier build should treat the JWT path as the material change and avoid attributing unrelated behavior to v2.2.0.

The release includes read, write, and TPC C benchmark tables. It does not describe them as improvements introduced by v2.2.0. Treat the numbers as a snapshot attached to the release, not as a comparison against 2.1.11.

The reported read mean multiplier is 0.88, while the write mean multiplier is 0.91. Individual tests vary more than those means suggest. covering_index_scan reports 17.63 for MySQL and 2.3 for Dolt, with a 0.13 multiplier. select_random_ranges reports 0.39 for MySQL and 0.65 for Dolt, with a 1.67 multiplier.

The write results show the same split. oltp_update_non_index reports 4.1 for MySQL and 3.02 for Dolt, while oltp_read_write reports 8.9 and 11.24. The TPC C table reports 95.75 TPS for MySQL and 52.65 TPS for Dolt at tpcc-scale-factor-1, a 1.82 multiplier in the table.

No hardware, data volume, server configuration, or run methodology appears in the notes. These figures are therefore useful for identifying query shapes to test, not for capacity planning. Reproduce the relevant scans, joins, and transaction mix on the deployment that will carry production load.

Inventory every service that mints JWT tokens for Dolt before changing the deployed tag. Validate initial authentication and renewal with authentication_dolt_jwt, then run the normal connection path through at least one worker replacement or job restart.

If the current deployment is already on 2.1.11, the notes identify no novel merged changes in v2.2.0. If it is on an earlier version, the authentication behavior is the stated compatibility risk. Keep performance qualification separate because the published tables do not claim a change between those versions.