Horizon Dashboard Troubleshooting Prompt
Diagnose OpenStack Horizon failures — login loops, blank panels, session/CSRF errors, slow page loads, and missing services — by tracing the Django app, Keystone, and API endpoint chain.
- Target user
- OpenStack operators supporting the web dashboard
- Difficulty
- Beginner
- Tools
- Claude, ChatGPT
The prompt
You are a senior OpenStack operator who has fixed every flavor of "Horizon is broken" — from login loops to blank Project pages to 504s under load. I will provide: - The symptom and any error shown to the user - Horizon deployment (Apache/mod_wsgi, uWSGI, container) - `local_settings.py` relevant bits (`OPENSTACK_HOST`, `*_URL`, `ALLOWED_HOSTS`, `SESSION_ENGINE`, `CACHES`) - Horizon/Apache logs and browser console errors - Whether the same action works via CLI (to isolate Horizon vs API) Your job: 1. **Isolate the layer** — first decide: is this Horizon (Django), Keystone (auth), or a downstream service API? The fast test: does the equivalent `openstack` CLI call as the same user work? If yes, it's Horizon/endpoint config; if no, it's a service problem. 2. **Login loops & CSRF** — check `ALLOWED_HOSTS`, `CSRF_TRUSTED_ORIGINS`, cookie `Secure`/domain mismatch behind a TLS proxy, clock skew breaking Fernet tokens, and a non-shared session backend across multiple Horizon replicas (use memcached/db, not local memory). 3. **Blank or missing panels** — a panel needs a working service endpoint in the Keystone catalog and the user's roles; trace `WEBROOT`, `OPENSTACK_KEYSTONE_URL`, and per-service `*_API_VERSIONS`. A single broken endpoint can blank a whole page. 4. **Static assets / 500s** — `collectstatic` not run, compressed-offline mismatch, wrong `WEBROOT`, mod_wsgi user can't read static dir. 5. **Performance** — N+1 API calls per page, no caching, Keystone token validation latency; recommend memcached for sessions and the token cache. 6. **Logs to read** — Horizon log, Apache/uwsgi error log, browser console/network tab, and the Keystone log for the matching request-id. 7. **Validation** — reproduce, fix, confirm across multiple replicas and after a fresh login. Output as: (a) a layer-isolation decision tree, (b) the specific `local_settings.py` keys to check per symptom, (c) exact log greps, (d) the fix with before/after config, (e) a hardening note (shared sessions, HTTPS cookies, memcached). Bias toward: isolating Horizon from the API first, shared-session correctness, reading the request-id across logs.
Run this prompt with AI
Test it, get an AI-improved version, or compare models — live in the Prompt Workspace. No copy-paste.
Related prompts
-
OpenStack Skyline Dashboard Gateway Deployment Design Prompt
Design a production Skyline (skyline-apiserver + skyline-console) deployment in Kolla-Ansible — Keystone integration, VIP/TLS through HAProxy, WebSocket console proxying, and a safe cutover path from Horizon — without breaking SSO or noVNC.
-
Keystone Token & Auth Failure Triage Prompt
Diagnose Keystone authentication and authorization failures (401/403, expired Fernet tokens, broken role assignments, federation/LDAP lookup errors) by tracing the auth request through token validation, catalog, and policy enforcement.
-
Cinder NetApp & NFS Driver Mount/Export Debug Prompt
Diagnose Cinder NFS/NetApp backend failures where volumes won't attach because of stale exports, mount option drift, or export-policy mismatches.
-
Keystone Fernet to JWS Token Provider Migration Prompt
Plan a safe cutover from the Fernet token provider to JWS (asymmetric) tokens across a multi-node Keystone deployment without invalidating live sessions.
More OpenStack prompts & error guides
Browse every OpenStack prompt and troubleshooting guide in one place.
Reading prompts? Get all 500 in one free PDF
500 battle-tested, copy-paste AI prompts engineered by a senior systems engineer — every one with fill-in placeholders and safety/back-out notes. Drop your email and it's yours.
- 500 prompts: Linux · Kubernetes · Terraform · OpenStack · GitLab · Docker · Monitoring · Incident Response
- Instant PDF download — yours free, forever
- Plus one practical AI-workflow email a week (no spam)
Single opt-in · unsubscribe anytime · no spam.