You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Managed agent updates by resource label, not only the host update-group
#69790
Large fleets already put environment / team / account on the Teleport resource as labels. Those labels drive RBAC. Managed Updates v2 does not use them.
Today an agent is in an update group only if the host was given that name (teleport-update enable --group). Auth and the Web UI can start, mark-done, or roll back a group. They cannot say “update every agent with env=staging.” If the host has no group name, it falls into the last group in autoupdate_config (the catch-all). Adding more groups in the config does not split those hosts.
I am posting this as an Idea, not a feature-request issue yet, to see who else wants the schedule (or a Web action) to follow resource labels.
What works today
Set the group on the host: teleport-update enable --group <name>.
Set the calendar in autoupdate_config (name, days, start_hour, wait_hours, canary_count).
From Auth / Web: start, mark-done, or roll back that name.
Install automation can write the same string it already uses for labels into --group. That is a host change, not a cluster selector.
Not “remove --group.” Keep the host name if people want it.
Would you use one of these?
Slice
What it is
1. Label match on the schedule
An autoupdate_config group can select agents by resource labels (env, team, …), the same way a role matches node_labels.
2. Web / tctl pick by label
Admin starts or pauses an update for “these labels,” without first renaming the host group.
3. Keep --group only
Labels stay for access. Updates stay a local string. Install tools keep writing both.
--group can stay as a pin, or as the fallback when no label rule matches.
Workaround today
On each host (or in the install playbook), run teleport-update enable --group with the same value you already use for a label. The cluster still does not read labels for updates.
Related issues
These help, but they still assume a name on the host:
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Large fleets already put environment / team / account on the Teleport resource as labels. Those labels drive RBAC. Managed Updates v2 does not use them.
Today an agent is in an update group only if the host was given that name (
teleport-update enable --group). Auth and the Web UI can start, mark-done, or roll back a group. They cannot say “update every agent withenv=staging.” If the host has no group name, it falls into the last group inautoupdate_config(the catch-all). Adding more groups in the config does not split those hosts.I am posting this as an Idea, not a feature-request issue yet, to see who else wants the schedule (or a Web action) to follow resource labels.
What works today
teleport-update enable --group <name>.autoupdate_config(name, days,start_hour,wait_hours,canary_count).--group. That is a host change, not a cluster selector.Docs: Managed Updates for Agents, Managed Updates v2 resources.
What we are asking the room
Not “remove
--group.” Keep the host name if people want it.Would you use one of these?
autoupdate_configgroup can select agents by resource labels (env,team, …), the same way a role matchesnode_labels.tctlpick by label--grouponly--groupcan stay as a pin, or as the fallback when no label rule matches.Workaround today
On each host (or in the install playbook), run
teleport-update enable --groupwith the same value you already use for a label. The cluster still does not read labels for updates.Related issues
These help, but they still assume a name on the host:
Questions
--groupon the host?--groupoverride a label rule, or should labels win?--groupfrom the same automation that sets labels” enough for you?Please comment even if the answer is “the host group is fine.” That is a useful signal.
All reactions