LiteLLM / LiteLLM

LiteLLM Ghost Models with Multiple Workers — Redis Pub/Sub Cache Invalidation Failure

Fix LiteLLM proxy where deleted models remain available on some workers after deletion via /model/delete API Includes evidence for LiteLLM troubleshooting demand.

Source-backed Last updated May 13, 2026 1 source Needs local verification

Category: LiteLLM
Error signature: Deleted models persist in other workers' local cache when running LiteLLM with --num_workers > 1
Quick fix: Compare the failing environment with a known working setup, then change one configuration value at a time.
Updated: May 13, 2026
Verification status: Source-backed
Evidence: 1 public source URL

Before you change production

This page includes public source URLs in the imported troubleshooting record. Compare those references with your version and environment before applying changes.

Reproduce the smallest failing action and save non-secret logs before changing configuration.
Check versions for LiteLLM, related SDKs, package managers, CI runners, and hosting providers.
Change one setting or dependency at a time, then rerun the same failing command or request.
Avoid destructive commands, credential rotation, billing changes, or security relaxations without a rollback plan.

What this error means

Deleted models persist in other workers' local cache when running LiteLLM with --num_workers > 1 is a LiteLLM failure pattern reported for developers trying to fix litellm proxy where deleted models remain available on some workers after deletion via /model/delete api. Based on the imported evidence, treat this as a tool-specific troubleshooting page rather than a generic API error.

Why this happens

LiteLLM issue #27852 reports that with --num_workers 4 and Redis cache, deleting a model added via /model/new API does not invalidate the local in-memory cache of other workers. Ghost models continue serving requests, creating inconsistent API behavior.

Common causes

Multi-worker LiteLLM deployments using Redis cache experience ghost models — models deleted via API continue to serve requests on some workers because Redis Pub/Sub doesn't invalidate local in-memory caches across all workers. This causes inconsistent behavior and stale model serving.
LiteLLM issue #27852 reports that with --num_workers 4 and Redis cache, deleting a model added via /model/new API does not invalidate the local in-memory cache of other workers. Ghost models continue serving requests, creating inconsistent API behavior.

Quick fixes

Confirm the exact error signature matches Deleted models persist in other workers' local cache when running LiteLLM with --num_workers > 1.
Check the LiteLLM account, local tool state, and provider configuration involved in the failing workflow.
Compare the failing environment with a known working setup, then change one configuration value at a time.

Platform/tool-specific checks

Verify the command, editor, extension, or API client that produced the error.
Compare local settings with CI, deployment, or editor-level settings when the error appears in only one environment.
Avoid deleting credentials, local model data, or project settings until the failing scope is clear.

Step-by-step troubleshooting

Capture the exact error message and the command, editor action, or request that triggered it.
Check whether the failure is account/auth, quota/rate, model/provider, local runtime, or deployment configuration.
Review the source evidence below and compare it with your environment.
Apply one change at a time and rerun the smallest failing action.
Keep the working fix documented for the team or deployment environment.

How to prevent it

Keep provider/tool configuration documented.
Record non-secret diagnostics such as tool version, provider name, model name, and command path.
Add a lightweight check before CI or production workflows depend on the tool.

Diagnostic flow for this page

Match Deleted models persist in other workers' local cache when running LiteLLM with --num_workers > 1 exactly before applying the quick fix.
Compare the failing environment with LiteLLM versions, account scope, provider settings, and deployment context.
Check the listed common causes in order, starting with the cause that best matches your logs.
Use the evidence status below to decide whether to confirm against public sources or official documentation.
Apply one reversible change, rerun the smallest failing action, and keep rollback notes.

FAQ

What does Deleted models persist in other workers' local cache when running LiteLLM with --num_workers > 1 mean in LiteLLM?

Deleted models persist in other workers' local cache when running LiteLLM with --num_workers > 1 indicates the workflow described by "LiteLLM Ghost Models with Multiple Workers — Redis Pub/Sub Cache Invalidation Failure" failed in a LiteLLM context. Start by confirming the exact signature and the component that emitted it.

What should I check first for Deleted models persist in other workers' local cache when running LiteLLM with --num_workers > 1?

Check whether this cause matches your environment: Multi-worker LiteLLM deployments using Redis cache experience ghost models — models deleted via API continue to serve requests on some workers because Redis Pub/Sub doesn't invalidate local in-memory caches across all workers. This causes inconsistent behavior and stale model serving.

Is the Deleted models persist in other workers' local cache when running LiteLLM with --num_workers > 1 guidance source-backed?

The "LiteLLM Ghost Models with Multiple Workers — Redis Pub/Sub Cache Invalidation Failure" page is labeled source-backed. It has 1 public source URL from github.com, so use that evidence status when deciding how much additional verification you need.

When should I avoid the quick fix for Deleted models persist in other workers' local cache when running LiteLLM with --num_workers > 1?

For "LiteLLM Ghost Models with Multiple Workers — Redis Pub/Sub Cache Invalidation Failure", avoid applying "Compare the failing environment with a known working setup, then change one configuration value at a time." when the error text differs, the affected version or provider is different, or the change would touch credentials, billing, DNS, production deploy settings, or irreversible data.

What should I collect before debugging Deleted models persist in other workers' local cache when running LiteLLM with --num_workers > 1?

For "LiteLLM Ghost Models with Multiple Workers — Redis Pub/Sub Cache Invalidation Failure", collect the failing LiteLLM command or request, non-secret configuration, versions, timestamps, request IDs when available, and the smallest reproduction that still triggers the signature.