Description
Describe the bug
In a dual stack cluster where the primary IP family is IPv6 services by default come up only with IPv6 addresses. This results in the backend pods being unable to find the redis service address because they are only looking for A records when the service is only causing AAAA records.
To Reproduce
Steps to reproduce the behavior:
- Go to '...'
- Click on '....'
- Scroll down to '....'
- See error
Expected behavior
The backend pod support looking for the redis service at an IPv6 only exposed address.
Screenshots
[2024-07-28 23:47:50.294] ERROR (RedisClient): [on-error] primary :: code: ENOTFOUND, msg: getaddrin
fo ENOTFOUND kubevious-redis
Environment Details:
- Kubernetes Distribution/Version [e.g. GKE v1.13.12-gke.25, OpenShift 4.3]
- Browser [e.g. chrome, safari]
- Version [e.g. v0.3]
Additional context
Add any other context about the problem here.
Changing the ipFamilies of the redis service from the chart to have either IPv4 only or also IPv4 fixes the lookup issue. The chart currently doesn't support this through values so a post patch process is required.
There is still an issue that Redis is unreachable still after chaing the Service family to IPv4 only.
Logs from backend/guard and probably other components that rely on redids.
[2024-07-28 23:52:34.821] ERROR (RedisClient): [on-error] primary :: code: ECONNREFUSED, msg: connec
t ECONNREFUSED 10.5.209.102:6379
[2024-07-28 23:52:36.225] ERROR (RedisClient): [on-error] primary :: code: ECONNREFUSED, msg: connec
t ECONNREFUSED 10.5.209.102:6379
[2024-07-28 23:52:37.127] ERROR (RedisClient): [on-error] primary :: code: ECONNREFUSED, msg: connec
t ECONNREFUSED 10.5.209.102:6379
I'm guessing the above is a fatal issue because I'm not seeing any other obvious errors and the UI has zero data.