pyats-linux-vmware（pyats linux vmware 命令行）

VMware ESXi Host Operations

Testbed Requirements

ESXi hosts must be defined in the pyATS testbed with os: linux (ESXi shell is Linux-based):

devices:
  esxi-host-01:
    os: linux
    type: esxi
    connections:
      cli:
        protocol: ssh
        ip: 10.0.0.100
        port: 22
    credentials:
      default:
        username: "%ENV{NETCLAW_USERNAME}"
        password: "%ENV{NETCLAW_PASSWORD}"

Note: SSH must be enabled on the ESXi host (disabled by default). Enable via vSphere Client → Host → Configure → System → Services → SSH → Start.

How to Call

All commands use pyats_run_linux_command:

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_run_linux_command '{"device_name":"esxi-host-01","command":"<command>"}'

Commands

VM Inventory

List All Virtual Machines

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_run_linux_command '{"device_name":"esxi-host-01","command":"vim-cmd vmsvc/getallvms"}'

Returns a table of all VMs on the ESXi host:

What to check:

• All expected VMs are present and registered
• VMX file paths reference expected datastores
• Virtual hardware versions are current (old versions may lack features)
• Guest OS type matches actual installed OS
• No orphaned or unexpected VMs

Snapshot Management

Get VM Snapshots

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_run_linux_command '{"device_name":"esxi-host-01","command":"vim-cmd vmsvc/snapshot.get 1"}'

The parameter is the Vmid from vim-cmd vmsvc/getallvms. Returns the snapshot tree for that VM:

• Snapshot name and description
• Snapshot ID
• Creation timestamp
• Whether it's the current snapshot (the "you are here" marker)
• Parent/child snapshot hierarchy

What to check:

• Stale snapshots — snapshots older than 72 hours degrade performance (growing delta disks)
• Snapshot chains — deep chains (>3 levels) cause I/O performance degradation
• Pre-change snapshots — verify a snapshot was taken before maintenance windows
• Disk space — snapshot delta files consume datastore space proportional to guest write activity

Workflows

1. VMware Inventory Audit

pyats_list_devices → identify ESXi hosts in testbed
→ vim-cmd vmsvc/getallvms per host → collect all VMs
→ Cross-reference with NetBox/Nautobot → flag undocumented VMs
→ Check virtual hardware versions → flag outdated versions
→ GAIT

2. Snapshot Health Check

pyats_list_devices → identify ESXi hosts
→ vim-cmd vmsvc/getallvms per host → get VM IDs
→ vim-cmd vmsvc/snapshot.get per VM → collect snapshot trees
→ Flag: snapshots > 72 hours old, chains > 3 deep, orphaned snapshots
→ Severity-sort (stale snapshots = WARNING, deep chains = CRITICAL)
→ GAIT

3. Pre-Change VM Verification

ServiceNow CR must be in Implement state
→ vim-cmd vmsvc/getallvms → identify target VM by name → get Vmid
→ vim-cmd vmsvc/snapshot.get {vmid} → verify pre-change snapshot exists
→ If no snapshot: STOP — snapshot must be taken before changes
→ Record baseline state in GAIT

4. Multi-Host VM Inventory

pyats_list_devices → identify all ESXi hosts
→ vim-cmd vmsvc/getallvms per host (parallel) → collect fleet-wide VM inventory
→ Aggregate: total VMs, VMs per host, hardware version distribution
→ Cross-reference with vCenter inventory (if available)
→ GAIT

Parallel Operations

Run VM inventory across multiple ESXi hosts concurrently:

# ESXi Host 1
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_run_linux_command '{"device_name":"esxi-host-01","command":"vim-cmd vmsvc/getallvms"}'

# ESXi Host 2
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_run_linux_command '{"device_name":"esxi-host-02","command":"vim-cmd vmsvc/getallvms"}'

# ESXi Host 3
PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_run_linux_command '{"device_name":"esxi-host-03","command":"vim-cmd vmsvc/getallvms"}'

For snapshot checks, first collect VM IDs per host, then run snapshot queries in parallel across all VMs.

vim-cmd Quick Reference

Note: Only getallvms and snapshot.get are covered by this skill. The reference above is for context — expand to other vim-cmd operations as needed.

Integration with Other Skills

Guardrails

• Always call pyats_list_devices first — verify ESXi hosts exist in the testbed
• Read-only operations only — getallvms and snapshot.get are read-only
• Never create/delete snapshots via this skill — snapshot creation/deletion impacts VM performance and requires a ServiceNow CR
• Never power-cycle VMs — vim-cmd vmsvc/power.* operations are destructive and require explicit authorization
• Validate Vmid before use — always get the Vmid from getallvms output; never guess or hardcode
• Flag stale snapshots — any snapshot > 72 hours should be flagged as WARNING; > 7 days as CRITICAL
• SSH access concerns — ESXi SSH is often disabled in production; document that it must be enabled for pyATS access
• Record in GAIT — every vim-cmd execution must be logged