Tarantool Cluster Manager 1.10
Release date: July 2, 2026
Latest release in the series: 1.10.0
In TCM 1.10.0, security and configuration capabilities have been extended. The key change is the ability to fully configure clusters, security policies, and user permissions via YAML already at the bootstrap stage, including LDAP, password policies, and access control list (ACL) rules; a new option lets you require a forced password change on first login after bootstrap.
Also added: YAML configuration validation that reports the
problematic field, creating users with a specified ID, an
etcd-restore alias for the restore command, and a ring chart of
TDB worker states on the Stateboard page. The output of
tcm generate-config has been synchronized with tcm.example.yml,
stale status handling and error color indication in Stateboard have
been improved, deleting a user now also deletes all associated ACL
permissions, and the addon field is now optional. Support for the
backup/restore commands has been discontinued.
The fixes affected the explorer, cluster management, storage, and configuration handling, including storage integrity protection: if unauthorized changes are detected, TCM is now locked and displays an error.
You can now fully configure clusters, security policies, and user
permissions via YAML configuration already at the bootstrap stage —
including LDAP parameters,
password policies, and access control
(ACL) rules in the initial-settings section.
Also added is the ability to require a forced password change on
first login after TCM bootstrap — a new
--security.require-bootstrap-password-change configuration flag
(environment variable TCM_SECURITY_REQUIRE_BOOTSTRAP_PASSWORD_CHANGE).
The flag works both with an automatically generated password and with
a password set via --security.bootstrap-password
(BOOTSTRAP_PASSWORD). When the flag is enabled, after logging in the
user is taken not to the main TCM interface but to the forced
password change screen, and access to all other functions is blocked
until it is completed; after changing the password, the user is
automatically returned to the main interface.
In addition, the initial-settings section now allows creating a
user with a specified ID.
Added TCM YAML configuration validation — on error, the message specifies the path to the problematic field.
The parameter values in tcm.example.yml and the output of the
tcm generate-config command have been synchronized: the generated
configuration now includes an initial-settings section showing all
possible parameters for clusters, users, roles, and the audit log.
The Stateboard page now has a ring chart of TDB worker states — it shows workers in the healthy, degraded, unhealthy, and stale statuses.
Improved handling of stale statuses: elements with the stale status now take priority over errors, are displayed as warnings, and no longer affect the error and warning counters of parent elements.
Changed color indication: if an instance has errors but its status is
running, the color in Stateboard changes from red to orange, and
the errors themselves are moved to a new Warnings tab.
The storage format of TCM object signature has changed. If, after upgrading from a previous version, an error like the following appears:
ERR fail to serve application err.code=GetErrCode err.message="failto get data\nfail to get data\nsignature \"rsapss\" not verified(missing) reason=failed to get data reason=failed to get data"
for backward compatibility, run the command ./tcm resign-data,
specifying the TCM storage connection parameters — etcd or a
Tarantool-based storage (the same way they are passed when starting
TCM):
- via the
--configoption with a path to a YAML file or a configuration string, for example:./tcm resign-data --config /path/to/config.yaml; - via environment variables (
TCM_STORAGE_TARANTOOL_*orTCM_STORAGE_ETCD_*).
If a YAML file is used to read the configuration, make sure it is accessible and correct.
Deleting a user now also deletes all ACL permissions associated with them.
Support for the backup/restore commands has been discontinued. For
backups, use the ./tcm config export command instead — TCM can then
be started using the exported configuration.
Added the etcd-restore alias for the restore command.
The addon field in the configuration is now optional.
- A
not found clustererror that occurred in the explorer when opening a space without an index. - An error in the migrations pie chart on Stateboard: the chart now only takes into account the last saved migration and is colored according to its status on a specific instance.
- An error that caused TCM not to find a cluster if the cluster was started on a different storage and restarted after being added to TCM and after TCM itself was restarted.
- An error when disabling a cluster.
- A validation error when creating entities in the
initial-settingssection. - Incorrect behavior with the draft key when running
resign-data. - An error that occurred when the configuration contained empty fields.
- An error in the access permission filter when editing a storage.
- An error that allowed TCM to continue operating with a corrupted storage; TCM is now locked and displays an error when unauthorized changes are detected.
- Cluster authentication: added a password decoder for securely passing storage credentials.