DevOps & cloud · Automation
hypervisor-guests
/hypervisor-guests
Use when shaping, sizing, backing up or updating guests on a hypervisor.
- Author
- vlat0101 ↗
- Repository
- agent-playbook ↗
- Licence
- MIT
- Last repository change
- 05.11.2026
Assessment and security
86/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability21 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content25 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
Deprecation
No signals detected. Repository changed on 05.11.2026, without references to retired models or APIs.
Licence
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Use when shaping, sizing, backing up or updating guests on a hypervisor.
Original author description
Use when shaping, sizing, backing up or updating guests on a hypervisor.
How it works
- One guest, one purpose
- Shape
- Sizing without lying to yourself
- Three mechanisms, three different protections
- Backups worth the disk
- Updates
- Host-level blind spots
- Done when
Original SKILL.md headings · 74 lines, 535 words
SKILL.md preview
---
name: hypervisor-guests
description: "Use when shaping, sizing, backing up or updating guests on a hypervisor."
version: 1.0.0
license: MIT
metadata:
hermes:
tags: [guide, devops]
---
# Hypervisor guests
The tenth guest should be as boring as the first.
## One guest, one purpose
A container running an app, a database and a cron job is three services in one
failure domain and one upgrade path. Split it. The cost is a little RAM; the benefit
is upgrading one thing without risking two.
Worth allowing: a product that genuinely is one unit (app plus its own database),
where splitting invents a network path for no reason.
## Shape
| Need | Shape |
|---|---|
| Single Linux service | Container |
| Docker host | VM, or a container with nesting enabled deliberately |
| Kernel module, custom kernel, GPU | VM |
| Non-Linux | VM |
| Migrate while running | VM with replication |
## Sizing without lying to yourself
- **vCPU: physical cores minus two, total.** Overcommit is not a tuning trick on a
hypervisor, it is a queue: everything slows together and nothing looks guilty.
- **RAM: allocation plus 10% ≤ physical.** Then look at pressure, not allocation.
- **Disk: alert at 80%.** At 95% writes are already being lost.
- **IOPS:** the number of guests on one spindle matters more than their CPU. Enough
writes on one drive produce IO wait that reads like a memory problem.
## Three mechanisms, three different protections
| Mechanism | Protects against | Does not protect against |
|---|---|---|
| Snapshot | A bad upgrade in the next hour | Disk failure, host failure, deletion |
| Backup to another disk/host | Corruption, deletion, bad upgrade | Host loss if stored on the same host |
| Replication | Host failure, minutes of loss | Accidental deletion — it replicates that |
| Off-site copy | Fire, theft, ransomware | Nothing. It is the last line |
**A snapshot is not a backup. A backup is not a backup until a restore has been done.**
## Backups worth the disk
1. One job, one schedule, exclusions written and reasoned.
2. Retention that covers "I noticed a week later".
3. Verified: the job's log shows success **and** one restore per quarter is real.
4. Monitored: freshness check alerts when the newest backup is older than expected. A
backup job that silently stopped is the most expensive silence in the estate.
## Updates
- Guest OS on a schedule, not when something breaks.
- Images updated one at a time, with the rebuild path in the repo so a bad update is
a rebuild rather than archaeology.
- Snapshot, update, verify the service, delete the snapshot.
- Decide which guests may reboot unattended, and write it down.
## Host-level blind spots
- The hypervisor needs patching, monitoring and a backup of its own config.
- Root pools fill from snapshots; check pool usage, not just guest disk.
- When many guests misbehave, the host's logs matter more than theirs.
- Out-of-band access is worth having on anything you cannot walk to.
## Done when
- Every guest has a documented purpose, shape and size.
- Every guest is in the backup job or explicitly excluded with a reason.
- One restore performed and dated.
- A freshness alert exists on backups.
Versions
The database has no other skills named hypervisor-guests.
Database snapshot: 13.11.2026 · package version b36413988d95 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 90macos-development-vm-workflow
Prepare and reset clean macOS development guests on Apple silicon. Use for restore-image compatibility, VM identity, installation, resources, OS-version testing, signing, privacy, and disposable macOS research guests. - 100overview
Domain router and cross-technology expert for server virtualization and hypervisor platforms. Covers Type-1 and Type-2 hypervisors, hardware-assisted virtualization, live migration, HA/FT, VM storage, VM networking, technology selection, and V2V migration. Use for… - 100citrix
Citrix Hypervisor (XenServer) expertise across all versions (6.x through 8.x, XCP-ng): Xen Type-1 hypervisor, dom0/domU architecture, XAPI toolstack, xe CLI, Storage Repositories (NFS, iSCSI, FC, GFS2), Open vSwitch networking, XenMotion live migration, pool management, HA with… - 90proxmox-operations
Operate Proxmox VE through the Proxmox AI platform — inspect nodes, clusters, guests, storage, backups and vRouters, and propose changes that a human confirms. Use whenever the user asks about their Proxmox infrastructure, their VMs or containers, node capacity or load, backup… - 100nix-config-update
Use when a version changes here: upgrading or updating a package or nixpkgs, bumping or pinning a flake input (a tag or commit, not a branch), or deploying the result to hosts, VMs, and MicroVM guests. - 86backup-and-recovery
Use when backing up machines or recovering lost data.