HOMELAB-246: Remove Anthropic Claude model from LiteLLM #26

Merged

aaron merged 39 commits from plane/HOMELAB-246-remove-anthropic-key into main 2026-03-23 03:36:59 +00:00

Conversation 1 Commits 39 Files changed 32 +1697 -39

aaron commented 2026-03-23 03:36:46 +00:00

Owner

Copy link

Summary

• Remove "smart" model (anthropic/claude-sonnet-4-20250514) from LiteLLM proxy config
• LiteLLM now routes only to local Ollama models (fast + code)
• Anthropic API key was from personal Claude subscription — should not be used for automated API calls

Test plan

• ArgoCD syncs updated LiteLLM config
• LiteLLM pod restarts and runs healthy
• Only fast and code models available via LiteLLM API

🤖 Generated with Claude Code

## Summary - Remove "smart" model (anthropic/claude-sonnet-4-20250514) from LiteLLM proxy config - LiteLLM now routes only to local Ollama models (fast + code) - Anthropic API key was from personal Claude subscription — should not be used for automated API calls ## Test plan - [ ] ArgoCD syncs updated LiteLLM config - [ ] LiteLLM pod restarts and runs healthy - [ ] Only fast and code models available via LiteLLM API 🤖 Generated with [Claude Code](https://claude.com/claude-code)

aaron added 39 commits 2026-03-23 03:36:46 +00:00

docs: add Langfuse v3 LLM observability design spec (HOMELAB-159) ... b866806aa6

Design spec for deploying Langfuse v3 as unified LLM observability
platform. Uses official Helm chart via ArgoCD, CNPG PostgreSQL,
Zitadel SSO, and phased rollout (core → LiteLLM → metrics → agent).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

plan: add Langfuse v3 implementation plan (HOMELAB-159) ... 3e1119791c

8-task plan: CNPG cluster, base values, Zitadel OIDC, ArgoCD apps,
Pangolin resource, deploy, LiteLLM integration, verification.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

feat: add Langfuse base Helm values (HOMELAB-159) ... 3e8ef6a386

Base values for official Langfuse chart v1.5.22: disable bundled
PostgreSQL (using CNPG), configure ClickHouse single-node, Valkey
standalone, MinIO with Longhorn storage, sign-up disabled.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

feat: add Zitadel OIDC application for Langfuse (HOMELAB-159) ... 5f629734c2

Creates Langfuse OIDC app in Zitadel project with callback URL at
/api/auth/callback/custom. Distributes client ID/secret as K8s
secret in langfuse namespace.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'docs: add Langfuse v3 design spec and implementation plan (HOMELAB-159)' (#7) from plane/HOMELAB-159-langfuse-observability into live ef0fd58d29

fix: increase Langfuse memory limits to prevent OOM (HOMELAB-159) ... 10028fac6f

Next.js app requires more memory than 256Mi. Bumped to 512Mi
request / 1Gi limit to prevent heap allocation failures.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'fix: increase Langfuse memory limits to prevent OOM (HOMELAB-159)' (#8) from plane/HOMELAB-159-langfuse-observability into live f584a712ca

fix: enable id_token_userinfo_assertion for Langfuse OIDC (HOMELAB-159) ... 43590caf98

Zitadel follows OIDC spec strictly — omits email from id_token when
access_token is also issued. This flag forces Zitadel to include
userinfo claims (email, name) in the id_token.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'fix: enable id_token_userinfo_assertion for Langfuse OIDC (HOMELAB-159)' (#9) from plane/HOMELAB-159-langfuse-observability into live a6a9014456

fix: enable sign-up for initial Langfuse account creation (HOMELAB-159) df1698cc86

Merge pull request 'fix: enable sign-up for Langfuse (HOMELAB-159)' (#11) from plane/HOMELAB-159-langfuse-observability into live c8c5ad4b78

docs: add Cilium network policy assessment and sample manifests (HOMELAB-141) ... 5d1732b6e0

Evaluate which intra-cluster traffic rules can be managed via
CiliumNetworkPolicy CRDs vs OPNSense. Add ADR-008 documenting the
layered security model and VLAN limitation (cilium/cilium#40247).
Include 4 sample manifests: default deny, DNS allow, monitoring
scrape allow, and system namespace protection.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

fix: correct CiliumClusterwideNetworkPolicy type in README (HOMELAB-141) ... 731ca37a41

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

chore: add .worktrees/ to .gitignore for parallel session isolation (HOMELAB-173) 3630d0659b

Merge pull request 'chore: add .worktrees/ to .gitignore (HOMELAB-173)' (#13) from plane/HOMELAB-173-worktree-setup into live b4434d6c42

HOMELAB-176: ci: add PR title format validation to CI pipeline ... 15cffd05ae

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

HOMELAB-176: ci: replace auto-version with PR-title-based release versioning ... 0b7552db9c

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'HOMELAB-176: ci: add commit convention enforcement and release versioning' (#14) from plane/HOMELAB-176-release-versioning into live cf2290132b

feat: add Langfuse OTLP exporter to OTel Collector + LiteLLM callback (HOMELAB-175) ... 1fa6588981

OTel Collector:
- otlphttp/langfuse exporter with retry/queue config
- filter/langfuse processor (OTTL): only agent-harness spans go to Langfuse
- traces/langfuse pipeline: OTel -> filter -> Langfuse
- NodePort service for VM 110 agent access

LiteLLM:
- Enable langfuse callback for LLM-level tracing

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'feat: add Langfuse OTLP exporter + LiteLLM callback (HOMELAB-175)' (#16) from plane/HOMELAB-175-langfuse-otel-litellm into live f8a519f544

fix: OTel Collector retry config — add time unit suffixes (HOMELAB-175) ... 6292deeac5

max_elapsed_time needs 's' suffix. Also add initial_interval and max_interval.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'fix: OTel retry config time units (HOMELAB-175)' (#17) from plane/HOMELAB-175-otel-retry-fix into live ac359b8626

fix: reduce Velero BSL sync frequency to cut Backblaze Class C usage (HOMELAB-179) ... d516c042bb

Velero's default 1-minute backupSyncPeriod and storeValidationFrequency
were generating ~3000-5000 Class C transactions/day against Backblaze B2,
hitting the 75% daily cap alert. Set both to 30m which is more than
adequate for a homelab with a single daily backup schedule.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'fix: reduce Velero BSL sync frequency to cut Backblaze Class C usage (HOMELAB-179)' (#18) from plane/HOMELAB-179-velero-sync-frequency into live 92a8ced398

feat: add resource limit alerts and fix Loki retention (HOMELAB-180) ... f5b472ce74

- Add Longhorn storage capacity alerts (disk 80%/90%, node storage 80%)
- Add observability stack alerts (Prometheus TSDB/WAL, Loki errors,
  Tempo compaction/ingester failures)
- Improve PVC alerts: tiered thresholds (75% warn, 90% crit) and
  predict_linear() for "filling in <4h" prediction
- Fix Loki retention: enable compactor retention_enabled (required for
  retention_period to actually enforce cleanup on 50Gi PVC)

Sub-tickets: HOMELAB-219, HOMELAB-220, HOMELAB-221, HOMELAB-222

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'feat: add resource limit alerts and fix Loki retention (HOMELAB-180)' (#19) from plane/HOMELAB-180-resource-limit-alerts into live 0e888bcce4

Merge pull request 'docs: Cilium network policy assessment (HOMELAB-141)' (#12) from plane/HOMELAB-141-cilium-network-policies into live d5215fef3d

HOMELAB-232: feat(atlantis): add base Helm values for Atlantis deployment ... 5d283a7594

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'HOMELAB-232: feat(atlantis): add base Helm values' (#20) from plane/HOMELAB-232-deploy-atlantis into live adfb3ff93e

HOMELAB-232: fix(atlantis): move workflows inside repoConfig block ... d78ed929ff

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'HOMELAB-232: fix(atlantis): move workflows inside repoConfig' (#21) from plane/HOMELAB-232-deploy-atlantis into live b126bf0bff

HOMELAB-188: feat(lab-director): add Helm chart for K8s deployment ... c23d1df4ff

Backend + frontend deployments, services, HTTPRoute with path splitting
(/api/* + /ws/* → backend, /* → frontend). Database URI from CNPG secret.
Configurable operator URL and hostname via prod overrides.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'HOMELAB-188: feat(lab-director): add Helm chart for K8s deployment' (#22) from plane/HOMELAB-188-k8s into live 0f59cb00c0

HOMELAB-233: feat(atlantis): add repo-level config for infra-core terraform ... 86d36b0489

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'HOMELAB-233: feat(atlantis): add repo-level config' (#23) from plane/HOMELAB-233-atlantis-webhooks into live 7d6de95b4a

HOMELAB-233: fix(forgejo): allow webhooks to private/internal IPs ... cf17952d5d

Required for Atlantis webhook delivery — default ALLOWED_HOST_LIST
blocks private IP ranges.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Merge pull request 'HOMELAB-233: fix(forgejo): allow webhooks to private IPs' (#24) from plane/HOMELAB-233-atlantis-webhooks into live 2c557c933d

HOMELAB-236: fix(litellm): update stale Ollama model names ... b1e475ea69

Model config referenced llama3.1:8b and qwen2.5-coder:14b but Ollama
has llama3.2:1b-instruct-q4_0 and qwen2.5-coder:7b. This blocked all
Ollama inference through LiteLLM, preventing the Langfuse callback from
capturing any generation data.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

HOMELAB-246: remove Anthropic Claude model from LiteLLM proxy config ... 1594f8f7fe

Remove the "smart" model (anthropic/claude-sonnet-4-20250514) from the
model list. LiteLLM now routes only to local Ollama models (fast + code).
The Anthropic API key was from a personal Claude subscription and should
not be used for automated API calls.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

aaron merged commit 78a31a993c into main 2026-03-23 03:36:59 +00:00

aaron deleted branch plane/HOMELAB-246-remove-anthropic-key 2026-03-23 03:36:59 +00:00

atlantis commented 2026-03-23 03:37:00 +00:00

Collaborator

Copy link

Ran Plan for dir: core/terraform/live/zitadel workspace: default

Plan Error

Show Output

running 'sh -c' '/atlantis-data/bin/terraform1.14.7 init -input=false -upgrade' in '/atlantis-data/repos/aaron/infra-core/26/default/core/terraform/live/zitadel': exit status 1
Initializing the backend...
╷
│ Error: Missing Required Value
│ 
│   on versions.tf line 4, in terraform:
│    4:   backend "s3" {
│ 
│ The attribute "bucket" is required by the backend.
│ 
│ Refer to the backend documentation for additional information which
│ attributes are required.
╵
╷
│ Error: Missing Required Value
│ 
│   on versions.tf line 4, in terraform:
│    4:   backend "s3" {
│ 
│ The attribute "key" is required by the backend.
│ 
│ Refer to the backend documentation for additional information which
│ attributes are required.
╵
╷
│ Error: Missing region value
│ 
│   on versions.tf line 4, in terraform:
│    4:   backend "s3" {
│ 
│ The "region" attribute or the "AWS_REGION" or "AWS_DEFAULT_REGION"
│ environment variables must be set.
╵

Ran Plan for dir: `core/terraform/live/zitadel` workspace: `default` **Plan Error** <details><summary>Show Output</summary> ``` running 'sh -c' '/atlantis-data/bin/terraform1.14.7 init -input=false -upgrade' in '/atlantis-data/repos/aaron/infra-core/26/default/core/terraform/live/zitadel': exit status 1 Initializing the backend... ╷ │ Error: Missing Required Value │ │ on versions.tf line 4, in terraform: │ 4: backend "s3" { │ │ The attribute "bucket" is required by the backend. │ │ Refer to the backend documentation for additional information which │ attributes are required. ╵ ╷ │ Error: Missing Required Value │ │ on versions.tf line 4, in terraform: │ 4: backend "s3" { │ │ The attribute "key" is required by the backend. │ │ Refer to the backend documentation for additional information which │ attributes are required. ╵ ╷ │ Error: Missing region value │ │ on versions.tf line 4, in terraform: │ 4: backend "s3" { │ │ The "region" attribute or the "AWS_REGION" or "AWS_DEFAULT_REGION" │ environment variables must be set. ╵ ``` </details>

aaron referenced this pull request from a commit 2026-03-23 03:37:01 +00:00

Merge pull request 'HOMELAB-246: Remove Anthropic Claude model from LiteLLM' (#26) from plane/HOMELAB-246-remove-anthropic-key into main

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels

No items

No labels

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

2 participants

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

aaron/infra-core!26

No description provided.