Conductor changelog

Conductor now keeps running when Docker or Core Hub is not reachable yet, and connects as soon as they are available

Added verification of module updates: a service that fails to start or stay up for 20 seconds is rolled back to its previous version (set UPDATE_VERIFY_SECONDS to change the time, 0 to disable the check)

Added rollback of Conductor’s own update on Linux, with progress logged to logs/conductor-self-update.log

Modules whose latest version cannot be read are now skipped without blocking the other updates

Docker Compose file comments and formatting are now preserved

Added diagnosis of failed update requests: Conductor checks the proxy, the internet connection, the MOLO17 services and Docker Hub, so Core Hub can explain who has to act and what to do

Added GET /connectivity endpoint to run the same check on demand

Proxies that require credentials (HTTP 407) are now reported as such

A repeated update request now waits for the running update instead of failing with "Another container action is in progress"

Improved reconnection to Core Hub

Conductor logs are now written to the logs folder of the Gluesync installation

Sign-outs and role changes now reach Conductor within 5 seconds instead of 30

Removed the proxy password from Conductor logs

Fixed detection of services whose Docker Compose labels are written as a map

Fixed version check error when the update service returns an error page

Fixed Chronos routing so it no longer intercepts all /api/ requests

Conductor now manages collect‑logs script updates, ensuring they are always aligned with the latest version.

Fixed Traefik conf to redirect http to https

Added new Grafana Dashboard version that includes the new metrics for per-entity source lag and cache lag

Added support for MOLO17’s Grafana image, including the new embedded Dashboard

Added support for Global Settings to configure legacy TLS connections

Conductor now waits for the CoreHub SDK token at startup instead of serving requests before authentication is ready;

CoreHub introspection failures return HTTP 503 (CoreHub unavailable) instead of a generic 401, so an unreachable CoreHub is no longer treated as a failed login;

Container action APIs return the upstream HTTP status and error details from failed image or registry calls instead of a generic 500;

Collect-logs WebDAV/FTP uploads default to a 60-minute timeout (was 10 minutes) and retry with backoff, so large log archives are less likely to fail mid-upload;

Breaking change: this version is only compatible with Core Hub 2.2.10+ onwards;

Added authentication via CoreHub token introspection with cookie and bearer token support;

Added role-based access control with SUPER_ADMIN, MANAGER, VIEWER, and EXTERNAL_MODULE roles;

Added route-level authorization guards on container, service, collect-logs, and migration endpoints;

Added sensitive data redaction in API responses for users without configuration-modification permissions;

Added readonly functionality for VIEWER role;

Expose ports 8765 on Core Hub for Query Forge service;

Improved collect logs file handling for Windows and Linux;

Better detection for .gz logs collection

Added core-hub heal support for old CPU

Updated base Node.js version to 24.16.0

Fixed CVE-related Dockerfile.windows configuration

Added suppport for .gz logs collection;

Lookback for logs raised from 3 days to 5 days;

Fixed an issue with Windows-based images retrieval;

Improved stability of collect logs script;

Better handling of release channels;

Fixed an issue where Backoffice maintenance mode errors were not forwarded correctly;

Updated fallback collect logs script for improved reliability;

Added a Watchdog guard to recover from stuck Conductor updates;

Added limits for collected log files and introduced a fallback collection script;

Added support for changelog retrieval;

Fixed Conductor restart in Windows server after its update

Added heal to Grafana configuration to allow visualization inside an iframe

Added support for collecting archived log files during log gathering

Improved log gathering for busy containers

Removed logs older than 30 days during log collection

Fixed log archiving on Windows

Enabled collect logs behind a proxy

Fixed log tar on Linux

Handled restart and update permissions when Conductor is installed on a different drive in Windows

Fixed log zipping on Windows

Logs are now collected by Conductor itself and not with a script anymore, with better error handling and reporting

Added optional proxy configuration to allow Conductor access to the internet

Artifacts published to MOLO17 repository;

Supported updates of all services of type Module

Supported migration to Gluesync 2.2

Allowed collect logs if Conductor is updating

Added DNS configuration for Conductor in docker-compose.yml, including autoheal support.

Added retry mechanism for container actions (add, remove, update agents and modules).

Changed "Restart Gluesync Instance" (Danger Zone) to perform a full stop and restart using docker compose down followed by docker compose up.

Updated logs uploader for Windows that now also includes docker stats

Updated docker client for more resilient reconnection in case of sock busy (particularly during cold starts of the Docker Engine on Windows)

Updated logs uploader for Windows new version with locked file check improvement;

Adoption of third-party services (Grafana, Prometheus, Traefik, Portainer) also if these are under the latest tag;

Fixed an issue that could cause conductor to return a mandatory update tag even if the module was up to date already;

Updated logs uploader for Windows and linux to new version with FTP improvement;

Updated collect logs script to improve diagnostics and reliability;

Stabilized Windows conductor update process and fixed issues that could prevent proper module updates;

Improved stability in service listing and latest-version retrieval within the Modules Manager menu;

Enhanced logic for retrieving the latest available module versions;

Fixed an issue affecting updates of third-party services.

Fixed an issue with FTP upload for Windows (file name encoding issue);

Fixed an issue that could cause the module to fail to update itself;

Updated logs uploader for Windows to version 1.5;

Automatically removed container_name and depends_on entries from Docker Compose configurations to improve portability and reduce conflicts.

Added explicit network configuration for Linux services to prevent rare cases where the SDK failed to reconnect.

Introduced TIMEZONE handling via .env, propagating the value into both core-hub and agents configurations.

Updated agents.json so Whisperer is correctly recognized and auto-adopted in Gluesync deployments.

Adopted third-party services (Grafana, Prometheus, Traefik, Portainer) into the deployment stack, allowing full control over version selection.

Fixed Dockerfile ARG handling to avoid build-time breakages and ensure proper support for the Docker helper image used during conductor restarts.

Fixed service version detection by resolving running containers via Docker Compose labels instead of relying on container names.

Fixes on logs uploader for Windows;

Used BASE_PATH variable in environments for Windows containers

Added support for restarting all services in containers when no IDs are specified.

Added agentId in list containers api response;

Fixed TAG env for agent nicknaming;

Removed container_name from docker compose file to avoid conflicts during updates;

Fixed service_id in docker compose file to use the container id instead of the agent id;

Fixed auto adopt removing the INITIAL_AGENT_ID handling;

Fixed empty object in docker compose file for deploy parameter;

Fixed create volume folder with container id instead of container_name;

Other minor fixes and improvements;

Updated logs uploader for Windows;

Updated logs uploader for Linux;

Added support for Windows containers on Windows Server 2019

Fixed network in some Windows instances

Added support for Windows containers on Windows Server 2022

Pruned images that have no running containers after an update;

Preserved architecture-specific suffixes in images during updates;

During its update, Conductor does not process requests and responds with a message and HTTP status code 503;

Updated the Node.js SDK to connect more fluently to Core Hub after the first connection;

Added support for specifying either a full URL or just a hostname in the GLUESYNC_HOST environment variable;

Improved version checking;

Fixed dependency issue for bootstrap-core-hub.json file under agents' management;

Auto adopt previously hand held services;

Removed depends_on in adopted services (even "adopted by hand");

Handled update of Conductor by doContainersAction update api;

Auto heal Conductor conf if some configuration of Conductor is wrong it will self heal (for now only logs file);

If Auto healed auto update Conductor version, no more latest in docker-file;

Edited get version api to notify when modules have mandatory updates to be applied.

Handled read of docker file configuration both in List Style and Map Style;

Handled updates only for services handled by Conductor;

Update with an empty array update all must accept the releaseChannel parameter;

With update all: update only containers with core-hub, agent, and module (no conductor);

List all containers except third party containers (grafana, prometheus, portainer, or traefik);

Fix current version in get latest;

Handled null version tag in get latest (which does not fires an actual update);

Updated Node.js SDK to the latest version;

Re-send license headers at every reconnection to the Core Hub;

Filter out containers without Id.

Added support for Collect logs API.

Added support for Agents TAG environment variable;

Fixed missing agent’s data folder.

Initial release