Diagnose Olares app install, crash, image pull, and reachability failures via olares-cli.
npx skills add https://github.com/beclab/Olares --skill olares-doctor
Please help me install this Agent Skill. Skill: olares-doctor Source: https://github.com/beclab/Olares/tree/main/cli/skills/olares-doctor Command: npx skills add https://github.com/beclab/Olares --skill olares-doctor
When an Olares app won't install, keeps crashing, can't pull its image, or shows as running but is unreachable, this Skill finds the root cause by correlating evidence from Market, Cluster, and Dashboard instead of guessing.
thirdleveldomain --force-dedupe) requires explicit approval.installing forever. The Skill checks whether another app is downloading (normal queue), inspects pod events for scheduling failures, and routes you to the right fix.Ask the AI to diagnose why my Olares app is stuck in installing and never reaches running.
High-intent search queries and answers about installing and using this skill.
Check `olares-cli market status` for the app state, then verify whether another app is downloading since app-service processes one download at a time. If nothing is downloading, inspect pod events and container statuses in the app's namespace for scheduling or pull failures.
Get the pod logs with `olares-cli cluster pod logs <ns>/<pod> -c <container>` and the `--previous` flag for the crashed instance's traceback. Common causes include missing env or secrets, service-link env collisions, and uid 1000 permission issues on userspace mounts.
The `running` state only proves entrance TCP reachability, not HTTP health. Check pod readiness, app logs for startup blockers, and the entrance proxy timeout, which defaults to 15 seconds via `options.apiTimeout`.
Run `olares-cli doctor images --unused` to list zero-reference images with reclaimable size. The command is read-only and only identifies candidates; cross-check running pods before pruning since tag bumps can make active images look unused.
Diagnosis is read-only by default and never restarts, deletes, or scales workloads automatically. The only mutation is `thirdleveldomain --force-dedupe`, which rewrites Application CRs and requires explicit user approval before running.
The rejection means the compute binding cannot fit available resources. Free resources or select different devices using `market resume --compute-binding`, and list operable devices with `olares-cli settings compute list`.
beclabRead and mutate Olares Settings areas with role-based access checks.
neon-law-foundationRemove unused Docker images, BuildKit cache, and detached volumes with storage reporting.
TzJ2006Parse sinfo output to map GPU resources across Slurm cluster partitions.
phuetzConsolidate kubectl, Helm, and ArgoCD workflows for Kubernetes cluster administration.