0

I'm running a Redis Cluster with 6 nodes (3 masters, 3 replicas) in Docker containers on Windows using bridge networking. Each container maps its port (e.g., 7010:7010, 7011:7011, etc.) so external clients can connect via 127.0.0.1:7010, 127.0.0.1:7011, etc. Current Behavior: When I create the cluster using internal Docker IPs (e.g., 192.168.144.2:7010), the cluster works perfectly for inter-node communication. However, when external clients connect, they receive MOVED redirections pointing to internal Docker IPs like 192.168.144.2:7011, which are unreachable from the host machine. What I've Tried:

Setting cluster-announce-ip 127.0.0.1 in redis.conf for all nodes - This causes all 127.0.0.1 nodes to show as "disconnected" in CLUSTER NODES because each container interprets 127.0.0.1 as itself, not the other containers. Creating cluster with 127.0.0.1 addresses - Fails with "Connection refused" because containers cannot reach each other via localhost. Applying announce settings after cluster creation via CONFIG SET - Results in mixed state where some nodes show 127.0.0.1 (disconnected) and others show internal IPs (connected).

Example nodes.conf showing the problem: 2f167b60 127.0.0.1:7010@17010 master - disconnected 0-5460 1b58373c 192.168.144.7:7013@17013 slave - connected e237ac08 127.0.0.1:7011@17011 myself,master - connected 5461-10922 Questions:

Is it possible to configure Redis Cluster so nodes communicate internally using Docker IPs (192.168.X.X) but announce 127.0.0.1 to external clients for redirections? Is there a correct order of operations (cluster creation → announce IP setting → restart/gossip) that makes this work? Are there alternative approaches for Docker bridge networking on Windows that don't require clients to use nodeAddressMap?

Environment:

Redis: 7.4.6 (redis-stack-server) Docker: Bridge network mode OS: Windows 11 Each node exposes ports in format PORT:PORT (e.g., 7010:7010, 17010:17010)

Goal: External clients should connect using only 127.0.0.1:7010-7015 without needing to know internal Docker IPs or configure nodeAddressMap.

Improve this question
3
  • check what MOVED redirects look like from a client, they’re probably pointing to internal docker IPs not 127.0.0.1? Commented Nov 15 at 15:03
  • Exactly, they are pointing internal docker IPs so they failed to write Commented Nov 16 at 7:51
  • yep, classic docker trap.. nodes announce their own container ip so MOVED redirects to unreachable internals,, set cluster-announce-ip to the docker gateway ip (same for all nodes) + correct cluster-announce-port per node and you're good! Commented Nov 18 at 22:10
Related questions
1
Cannot Connect to Redis Cluster inside Docker from Client
1
Can not connect SpringBoot to redis-cluster on docker
122
When to use Docker-Compose and when to use Docker-Swarm
Load 3 more related questions Show fewer related questions

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.