Bundle: cluster-manager
A cluster manager's knowledge instance - documentation, tickets, operational memory and, optionally, ServiceNow, Slurm, HTCondor, node health and code.
| Field | Value |
|---|---|
| Version | 0.3.0 |
| Extends | nothing |
| Okg versions it supports | >=0.6.0,<0.7 |
Packs
- Default:
docsite,tickets,ops-record - Optional:
servicenow,slurm,htcondor,git-code,node-health,ticket-triage
Default packs install with the bundle. Add an optional pack with --with <pack> at init, or archi add <pack> later.
archi init myteam --bundle cluster-manager
Answers the bundle fixes
| Id | Value |
|---|---|
docsite_urls |
https://slurm.schedmd.com/documentation.html,https://htcondor.readthedocs.io/en/latest/ |
Chat preset
| Setting | Value |
|---|---|
site_name |
Cluster operations chat |
default_model |
llama3.1:8b |
Suggested questions:
- How do I drain a node in Slurm, and what does the documentation say about it?
- Which tickets mention a node that is currently down?
- Have we seen this error before, and what fixed it?
Automation level bounds
The bundle sets no bound. Each automation pack's own limit applies.