Elasticache Redis (cluster mode enabled) NXDOMAIN error

Question

Hello!

I've started new Elasticache Redis cluster (cluster mode enabled) with 1 shard and 2 nodes in it, cache.t4g.micro, using terraform module by CloudPosse.

After couple of weeks second node lost it's DNS record.

First node is ok:
```
host zzz-0001-001.zzz.yyy.euw1.cache.amazonaws.com

zzz-0001-001.zzz.yyy.euw1.cache.amazonaws.com has IPv6 address aaaa:aaa:aaaa:aaa:aaaa:aaaa:aaaa:aaaa
```

But second node returns dns NXDOMAIN error:
```
host zzz-0001-002.zzz.yyy.euw1.cache.amazonaws.com

zzz-0001-001.zzz.yyy.euw1.cache.amazonaws.com not found: 3(NXDOMAIN)
```

It started only after a couple of weeks after initial start.

I see this error after connect via teleport (via redis-cli the same error is shown):
```
tsh db connect --db-user my-user zzz

localhost:65293> keys *
(error) ERR Teleport: dial tcp: lookup zzz-0001-002.zzz.yyy.euw1.cache.amazonaws.com on [fdf3:aaaa:aaaa::a]:53: no such host
(0.76s)
```

Accepted Answer

Answer from AWS support:

> On 2024-03-26 at 18:07:00 UTC,  ElastiCache applied patching update `elasticache-20240225-arm` for your replication group named qqq in yyy region. Due to a bug , ipv6 address for your node `qqq-0001-001` was removed from the endpoint.We have since fixed the issue for your cluster and ipv6 address has been updated for your replication group.

Answer

This is not the answer you expect, but when cluster mode is enabled, the Configuration endpoint is used, so I don't think you would use the per-node endpoint directly.

https://docs.aws.amazon.com/AmazonElastiCache/latest/red-ug/Endpoints.html

Of course, the per-node endpoint can be named resolved.
If the node's endpoint cannot resolve the name, it may be that the node is currently unavailable. Check the node's status in the management console.

Elasticache Redis (cluster mode enabled) NXDOMAIN error

Contenido relevante