Pacemaker MCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| pcs_cluster_statusA | Displays high-level cluster state using 'pcs cluster status'. Useful for quick health checks: shows whether Pacemaker is running, node membership, fencing/stonith state, and manager health. |
| pcs_node_statusB | Shows node states using 'pcs status nodes'. Lists each node and whether it is online/offline, standby, maintenance, or otherwise unavailable. |
| pcs_resource_listC | Shows full resource configuration using 'pcs resource config'. Includes agents, parameters, operations, meta-attributes, groups, and ordering/colocation metadata for troubleshooting and audits. |
| pcs_statusA | Overall cluster status via 'pcs status': summarizes nodes, resources, failures, and constraints. |
| pcs_status_resourcesA | Resource runtime state via 'pcs status resources': shows which resources are started/stopped and on which nodes. |
| pcs_status_xmlA | Cluster status in XML via 'pcs status xml' for programmatic parsing and deeper diagnostics. |
| pcs_status_fullA | Full cluster configuration and runtime details via 'pcs status --full'. Includes nodes, resources, failures, options, constraints, and history useful for audits and deep troubleshooting. |
| pcs_cluster_startC | Start the cluster stack on the node via 'pcs cluster start'. |
| pcs_cluster_stopC | Stop the cluster stack on the node via 'pcs cluster stop'. |
| pcs_node_standbyB | Put a node in standby (no resources run there) via 'pcs node standby []'. |
| pcs_node_unstandbyB | Bring a node out of standby via 'pcs node unstandby []'. |
| pcs_constraint_listA | List all constraints via 'pcs constraint config' including ordering and colocation. |
| pcs_property_listA | List cluster properties via 'pcs property list --all' for tuning and defaults. |
| pcs_stonith_showC | Show stonith/fencing devices via 'pcs stonith show' including configuration and parameters. |
| pcs_resource_enableB | Enable a resource or all resources via 'pcs resource enable ' or '--all'. |
| pcs_resource_disableB | Disable a resource or all resources via 'pcs resource disable ' or '--all'. |
| pcs_resource_moveB | Create a temporary location constraint via 'pcs resource move []'. |
| pcs_resource_clearC | Clear temporary constraints and failures for a resource via 'pcs resource clear '. |
| pcs_resource_cleanupC | Run cleanup on a resource (or all) via 'pcs resource cleanup []'. |
| pcs_resource_banB | Ban a resource from a node via 'pcs resource ban [lifetime]'. |
| pcs_resource_unbanC | Remove a ban for a resource via 'pcs resource unban []'. |
| pcs_resource_restartB | Restart a resource via 'pcs resource restart ' or all resources if omitted (agent-dependent). |
| pcs_resource_configA | Show the configured definition for a specific resource via 'pcs resource config '. Includes agent, parameters, operations, and meta-attributes for that resource only. |
| pcs_logs_tailA | Tail last N lines from one or more log file paths. Skips missing files. |
| pcs_logs_commonC | Retrieve common Pacemaker/Corosync logs and optional journal snippets. |
| pcs_logs_journalctlB | Read journal logs via journalctl; defaults to pacemaker and corosync units. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 26 tools
There are many overlapping status/read tools: pcs_status, pcs_status_full, pcs_status_xml, pcs_cluster_status, pcs_status_resources, pcs_resource_list, and pcs_resource_config all expose related cluster or resource information. Agents would have difficulty choosing between them without reading descriptions closely, and several boundaries are unclear.
All tools share a pcs_ prefix, but the naming pattern is mixed: some use object_verb (pcs_cluster_start, pcs_resource_enable), others use verb-like object suffixes (pcs_status_resources, pcs_status_xml), and list/show are used inconsistently. It is readable but not predictable enough for reliable selection.
26 tools is excessive for the apparent scope, especially given redundant status variants and multiple log retrieval tools that could be consolidated. Many of the tools overlap heavily, so the count feels bloated rather than well-scoped.
The server covers inspection and some operational actions like enabling, disabling, moving, and banning resources, but it lacks configuration CRUD: no resource create/delete/update, no constraint creation/deletion, no property changes, and no stonith device management. This creates significant gaps for a tool set claiming to manage a Pacemaker cluster.