[Bug]: Cannot connect to AWS Redis cluster over SSM

[yassinecc]

Preconditions

I had previously set up SSM connection through a bastion instance, and I am able to connect with redis-cli

Steps to reproduce

• Create a Redis cluster (cluster-mode enabled) on AWS
• Allow bastion host access on cluster security group over port 6379
• Retrieve cluster configuration endpoint from AWS console
• Start SSM session with

aws ssm start-session --target $bastion_instance_id --profile $profile \
    --document-name AWS-StartPortForwardingSessionToRemoteHost \
    --parameters host="$cluster_endpoint",portNumber="6379",localPortNumber="6370"

• Create a new database connection on Redis Insight, using host 127.0.0.1, port 6370

Actual behavior
I see a "Could not connect to 127.0.0.1:6370, please check the connection details." error message

Expected behavior
Connection should be successful

Screenshots

Additional context
OS: Ubuntu 22
Redis server: AWS Elasticache Redis 7.1
Redis Insight: v2.48
Redis CLI: v6.0.16

[ViktarStarastsenka]

hi @yassinecc , can you please check that all the cluster nodes are available?
Redis Insight needs all the cluster nodes to be available to connect to a clustered database.
Though redis-cli connects to only one node.

A quick check that you can do is try to redirect (e.g. when there is no such key in this node). If other nodes are not available, then redis-cli will get a connection timeout as well

[ViktarStarastsenka]

@yassinecc , try this: #3429 (comment)

[yassinecc]

Hello @ViktarStarastsenka, connection fails even with only one node in the cluster and without using TLS authentication, but I will try building RedisInsight with the longer timeout and let you know

[ViktarStarastsenka]

please re-open the ticket if the solution above does not fix the issue.

[pdeveltere]

@ViktarStarastsenka This is still an issue for us as well. Cannot connect to the redis cluster with the same steps to reproduce as described here.

[pdeveltere]

@ViktarStarastsenka the only difference with the described steps is that our redis instance is a serverless instance, on AWS elasticache.

A call like redis-cli --tls is working

[pdeveltere]

@ViktarStarastsenka I cloned the repo locally and found the issue. The problem is that for serverless instances, we should not connect in cluster mode, since AWS takes care of that itself.

So when I disable the following line it works:
https://github.com/RedisInsight/RedisInsight/blob/8e486395d0d5caf154366909eb1d7995e7829458/redisinsight/api/src/modules/database/providers/database.factory.ts#L61-L63

I am trying to figure out a way to detect that the server is an AWS serverless instance, but not luck so far. Perhaps you could also include an option to override either connecting in cluster mode vs standalone?

[ViktarStarastsenka]

@pdeveltere , thank you for looking into this. I'll see what we can do here for November-December releases.

[ViktarStarastsenka]

An update: We have a ticket for this issue in our backlog, but it has not been prioritized yet.

[Mikephii]

Any update?

[ViktarStarastsenka]

this will be delivered with the March release (I think end of March)

[yassinecc]

FYI the original issue still persists on the last version 2.68.0