Skip to main content
Glama

Run OS patches

run_os_patches
Destructive

Start the rolling OS security patch of an Ankra-provisioned cluster: one host at a time, never two. Each node is checked first (nothing pending means nothing is touched), then cordoned and drained through a peer control plane with PodDisruptionBudgets honoured and never forced, upgraded with apt (the held-back systemd, udev, netplan, container-runtime and packet-filter packages included; packages are never removed and Kubernetes packages are never moved), rebooted only when the update asks for it, and returned to service only once it is Ready, its etcd member is healthy (control planes, which go one at a time and only with quorum to spare), its workloads are back and the cluster's agent has checked in. A failed drain or host stops the whole run. Nodes carrying single-replica workloads with volumes, or in a database node pool, are skipped and reported unless include_stateful_nodes is true; a cluster's only control plane is never patched live. When the cluster's OS patch policy is enabled, each host the run patches or finds up to date gets Ankra's update hold (apply_node_os_policy), and a skipped stateful node or only control plane has it removed, so every host is patched by either this lane or unattended-upgrades. The bastion goes last; on clusters where it is the NAT gateway its reboot briefly cuts the nodes' internet egress, and its NAT and SSH guard are re-asserted afterwards. The run starts only inside the cluster's maintenance window unless outside_window is true. Run plan_os_patches first to see what will happen, and follow the run with get_os_patch_status.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "run_os_patches"
  }
}

Input Schema

NameRequiredDescriptionDefault
cluster_idNoUUID of the cluster. Optional when a cluster is selected.
cluster_nameNoExact name of the cluster. Optional when a cluster is selected.
outside_windowNoStart now even though the cluster's maintenance window is closed. Only when the user explicitly wants that.
include_bastionNoPatch the bastion as the last step. Defaults to true.
include_stateful_nodesNoAlso drain and patch nodes carrying single-replica workloads with volumes or belonging to a database node pool. Defaults to the cluster policy (off). Only when the user accepts that those workloads go down while their node is patched.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Added

TDQS

A4.7/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations carry destructiveHint=true and readOnlyHint=false, but the description goes well beyond them: serial one-host-at-a-time execution, cordon/drain with PDBs never forced, explicit package holds, reboot-only-if-required, readiness and etcd quorum gates, abort-on-failure semantics, skipped stateful nodes, the single-control-plane exclusion, and the bastion's NAT egress interruption. That is exactly the behavioral context an agent needs before invoking a destructive fleet operation.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

It is front-loaded correctly with the action and contains no filler, but it is one dense run-on paragraph of roughly 200 words where a short structured list of guarantees and side effects would scan far faster. Every clause is informative, yet the packing into continuous prose makes key constraints (abort-on-failure, control-plane quorum) easy to miss.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a destructive, multi-phase, non-idempotent tool with no output schema, the description covers prerequisites (maintenance window, planning first), execution semantics, skip rules, failure behavior, and post-run verification. Nothing material about invoking it safely is left unstated.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the baseline is 3, but the description adds real consequence-level meaning: include_stateful_nodes is tied to skipped single-replica/database nodes and workload downtime, outside_window to the maintenance-window bypass, and include_bastion to the last-step ordering and NAT-gateway side effect. It does not restate types or defaults, it explains what each flag changes.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The first clause names a specific verb and resource: 'Start the rolling OS security patch of an Ankra-provisioned cluster.' It is immediately distinguishable from plan_os_patches (planning), get_os_patch_status (status) and apply_node_os_policy (per-host hold), which are all named in the text.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Explicit routing is given: 'Run plan_os_patches first to see what will happen, and follow the run with get_os_patch_status.' The maintenance-window precondition and the outside_window exception are stated, plus the condition under which stateful nodes are included, so the agent knows both when to use this and when not to.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.