CROSSSLOT Redis Error Causing Prompt Creation Failure with Azure Cache for Redis in Cluster Mode (Self host in openshift)

[coderCyFl]

SELF HOST LANGFUSE IN OPENSHIFT KEEP GETTING CREATE PROMPT FAIL

Title: CROSSSLOT Redis Error Causing Prompt Creation Failure with Azure Managed for Redis in Cluster Mode

Description:

I'm experiencing prompt creation failures in my self-hosted Langfuse deployment using Azure Cache for Redis. The operation fails with a CROSSSLOT error.

Environment:

Langfuse Version: v3.140.0
Redis Provider: Azure Managed Redis

Deployment Method: [Docker/Kubernetes/other]
Error Logs:

{"level":"warn","message":"Redis connection error: CROSSSLOT Keys in request don't hash to the same slot (context='', command='eval', original-slot='8819', wrong-slot='12233', first-key='bull:entity-change-queue:wait', violating-key='bull:entity-change-queue:paused')","timestamp":"2026-02-03T06:45:26.680Z"}

{"level":"error","message":"Failed to queue entity change event for prompt 2b83c560-aa17-4da1-b902-c2e3eadab14e for project cml68gogr0005np07ufkc3tvy: ReplyError: CROSSSLOT Keys in request don't hash to the same slot (context='', command='eval', original-slot='8819', wrong-slot='12233', first-key='bull:entity-change-queue:wait', violating-key='bull:entity-change-queue:paused')","timestamp":"2026-02-03T06:45:26.682Z"}

{"code":"INTERNAL_SERVER_ERROR","level":"error","message":"tRPC route failed on prompts.create: Internal error. Please check error logs in your self-hosted deployment.","name":"TRPCError","timestamp":"2026-02-03T06:45:26.682Z"}
Additional Warning:

IMPORTANT! Eviction policy is volatile-lru. It should be "noeviction"
Issue:

Prompt creation fails because:

Azure Redis is running in cluster mode, causing CROSSSLOT errors when Langfuse queue operations try to access keys that hash to different slots1
Redis eviction policy is set to volatile-lru instead of the required noeviction1
Questions:

1. What are the correct environment variables to configure Langfuse for Azure Redis cluster mode?1
2. Should I set REDIS_CLUSTER_ENABLED=true and provide cluster configuration?1
3. Do I need to configure LANGFUSE_INGESTION_QUEUE_SHARD_COUNT and LANGFUSE_TRACE_UPSERT_QUEUE_SHARD_COUNT for queue sharding?(https://langfuse.com/self-hosting/configuration/scaling)
4. How do I change the Azure Redis eviction policy to noeviction?

Langfuse Cloud or Self-Hosted?

Self-Hosted

If Self-Hosted

v3.140.0

If Langfuse Cloud

No response

SDK and integration versions

Langfuse Version: latest
Infrastructure:

• Database: Azure Database for PostgreSQL
• Blob Storage: Azure Blob Container
• Cache: Azure Managed Redis (Clustering Policy
  OSSCluster)

Pre-Submission Checklist

• I have checked for existing issues/discussions and consulted Langfuse AI.

[inkeep-ai-assistant[bot]]

Based on the documentation, your CROSSSLOT errors are occurring because Azure Cache for Redis in cluster mode requires specific configuration that isn't currently set(1)(2).

Required Configuration for Azure Redis Cluster Mode

To resolve the CROSSSLOT errors, you need to enable Redis cluster mode with these environment variables(1):

REDIS_CLUSTER_ENABLED=true
REDIS_CLUSTER_NODES=your-cluster-endpoint:6379
REDIS_AUTH=your-redis-password
REDIS_TLS_ENABLED=true
REDIS_TLS_SERVERNAME=your-azure-redis-endpoint

(1)

For Azure Cache for Redis specifically, the configuration should look like(1):

REDIS_CLUSTER_ENABLED=true
REDIS_CLUSTER_NODES=clustercfg.my-redis-cluster.abc123.cache.amazonaws.com:6379
REDIS_AUTH=your-auth-token
REDIS_TLS_ENABLED=true
REDIS_TLS_SERVERNAME=clustercfg.my-redis-cluster.abc123.cache.amazonaws.com

(1)

Replace the endpoint with your Azure Redis cluster endpoint(1).

Queue Sharding Configuration

To distribute queue operations across cluster nodes and avoid CROSSSLOT errors, you should configure queue sharding(3):

Set LANGFUSE_INGESTION_QUEUE_SHARD_COUNT and LANGFUSE_TRACE_UPSERT_QUEUE_SHARD_COUNT to a value greater than 1(3). The documentation recommends a value that is approximately 2-3 times the number of shards in your Redis cluster(3).

For example, if you have 3 Redis shards, set these to 6-9(3):

LANGFUSE_INGESTION_QUEUE_SHARD_COUNT=6
LANGFUSE_TRACE_UPSERT_QUEUE_SHARD_COUNT=6

(3)

⚠️ Important: Once you've sharded your queues, do not reduce the number of shards(3). You should also scale LANGFUSE_INGESTION_QUEUE_PROCESSING_CONCURRENCY and LANGFUSE_TRACE_UPSERT_WORKER_CONCURRENCY accordingly, as these count per shard(3). The default target is a concurrency of 20 per worker, so if you have 10 queue-shards, set it to 2(3).

Redis Eviction Policy

You must set the maxmemory-policy parameter to noeviction to ensure queue jobs are not evicted from the cache(1). For Azure Cache for Redis, you'll need to configure this through the Azure portal or CLI in your Redis instance settings(1).

Additional Cluster Mode Requirements

When using Redis cluster mode(1):

• Ensure all cluster nodes are accessible from your Langfuse containers(1)
• Use the same authentication credentials across all cluster nodes(1)
• Monitor cluster health and handle node failures appropriately(1)
• Set maxmemory-policy=noeviction on all cluster nodes(1)

📚 Sources:

1. Cache (Redis/Valkey)
2. Error when using ElastiCache Serverless: CROSSSLOT Keys in request don't hash to the same slot
3. Scaling

---

Have another question? Just tag @inkeep.

[jannikmaierhoefer]

Hi @coderCyFl, was inkeep able to help you here?

[coderCyFl]

No . May i know there is any official support from langfuse side ?

[jannikmaierhoefer]

Looping in @Steffen911 to have a look.

[Steffen911]

@coderCyFl The inkeep answer seems to be on point for the questions you've asked.

What are the correct environment variables to configure Langfuse for Azure Redis cluster mode?1

The ones that are listed on the documentation page for Redis cluster mode.

Should I set REDIS_CLUSTER_ENABLED=true and provide cluster configuration?1

Yes.

Do I need to configure LANGFUSE_INGESTION_QUEUE_SHARD_COUNT and LANGFUSE_TRACE_UPSERT_QUEUE_SHARD_COUNT for queue sharding?(https://langfuse.com/self-hosting/configuration/scaling)

No.

How do I change the Azure Redis eviction policy to noeviction?

See the Inkeep response or the corresponding Azure documentation.

[coderCyFl]

Hi @Steffen911. Thanks for your update, all the necessary configuration i have been already set up. For REDIS_CLUSTER_ENABLED and other . Currently i found out keep showing this error :

Since i am using Azure Managed Redis 7

May i know the issue is related with eviction policy.

[coderCyFl]

I don't see the CROSSSLOT Redis Error anymore however when i create the prompt keep stuck at loading . May i know the root cause is related with Azure Managed Redis Policy Setting or what ? @jannikmaierhoefer @yarikoptic @deejay1 @Steffen911 .

[coderCyFl]

@Steffen911 @yarikoptic. Any idea? Many Thanks

[Steffen911]

@coderCyFl The eviction policy should be unproblematic for the issue you're observing. In addition, failing to refresh the slots cache is mostly unproblematic and resolves eventually.

Can you share which call you're trying to make and what errors you observe on the client and in the web container for this call? I.e. what is the failure that you observe.

In addition, I would ask to stop tagging people in this thread, especially non-maintainers.