Skip to content

v1.33

Latest
Latest

Choose a tag to compare

View all tags
@hermanekt hermanekt released this 31 Jul 11:03
v1.33
2ac1ff4
This commit was signed with the committer’s verified signature.
hermanekt Tomas Hermanek
GPG key ID: 36A739221F03A20A
Verified
Learn about vigilant mode.

v1.33 - 2026-07-31

Patch release. One admin-portal bug fixed; no new features. The v2.0 feature train (LDAP/SAML SSO, plugin loader, tool-level audit log, SIEM forwarder, /metrics) continues on the release/v2.0.0 branch.

Fixed

  • Endless "Configuration changes are pending. The MCP server must be restarted" loop (#61, reported by @G0nz0uk). Since the resilience change that skips an invalid [zabbix.X] section at boot instead of taking the whole MCP down, a section that fails validation (bad URL, missing token, unresolvable ${ENV_VAR}) is invisible to the runtime but still present in config.toml. The admin portal interpreted that permanent gap as "in config but not live yet" on two surfaces: the dashboard showed the server as pending, and the /servers page re-armed the global restart banner on every page render - so even after a restart cleared it, the first visit to /servers raised it again. Operators restarted in a loop; restarting can never load a section that fails validation.

    The config loader now records skipped sections with their validation reasons, and the admin portal says what is actually wrong: the /servers card gets a red "Config error" badge, the exact validation message, and an explicit "restarting will not load it - fix the section" note; the dashboard card shows "config error" with the reason in the tooltip; /api/server-status reports status: "config_error". Skipped sections are excluded from restart-drift detection, so the banner no longer resurrects itself.

Dashboard - skipped section now says what is wrong instead of "pending":

Dashboard with config error label on the skipped server

Zabbix Servers page - red badge + the exact validation error + what to do:

Servers page card with Config error badge and validation reason

Thanks

Thanks to @G0nz0uk for reporting #61 with screenshots and for patiently confirming that the restart loop survived every restart - the "restart does not clear it" detail is what pointed away from state handling and straight at the drift detector.

Support policy

Versions older than 1.30 no longer receive fixes (see SECURITY.md). Upgrade with ./deploy/install.sh update - config and tokens are preserved.

Verified

  • Live demo container reproducing the reporter's setup (valid + invalid [zabbix.X] section)
  • CRUD smoke 255/261 OK (1 environment error = graph_render needs frontend creds, 5 expected fixture-related skips)
  • Installer matrix 18/18 PASS across alma 8/9/10, debian 12/13, fedora, oracle 8/9/10, rhel 8/9/10, suse 15, ubuntu 22/24, amazon 2023, minimal

Upgrade

sudo /opt/zabbix-mcp/venv/bin/pip install --upgrade zabbix-mcp-server==1.33
sudo systemctl restart zabbix-mcp-server

or with the installer:

curl -fsSL https://raw.githubusercontent.com/initMAX/zabbix-mcp-server/v1.33/deploy/install.sh | sudo bash -s -- update
Assets 4
Loading