Hey Gabriel,
Thanks so much for pointing out the -c option. You’re absolutely correct that it’s needed in order for RedisCLI to know we’re using a cluster as opposed to a standalone instance.
Can the pods reach one another, i.e curl{gateway_url}:{port}/hello and/or list keys (keys *) after adding a random key? This will give us whether the pods can communicate with each other or not. This will also help us understand if your Redis connection is storing proper values. I noticed your earlier comment about different/ same namespaces but I just want to reconfirm that above again ^.
- Please post the
keys *from the master. - Also what is returned from your {gateway_url}:{port}/hello endpoint?
I also notice from the gateway you provided that you have addrs inside storage and then you have redis_addrs. Can you confirm you deleted the redis_addrs portion entirely?
Alternatively, i’d also recommend isolating the problem a bit further. At the current moment, we’re not entirely sure whether it’s actually the gateway or Redis that’s the issue (obviously the connection isn’t being made but isolating it would help us get closer to the solution). One way of doing this is by scraping away the cluster entirely and trying it with a standalone Redis instance (this takes away a bit of complications away). Assuming this works properly then our takeaway is that the cluster was incorrectly configured. If the connection is still not made properly, then the takeaway is that Tyk is at fault.
Additionally are you able to share a public repo of all the files you have. Alongside that can you share the steps (commands included) that you’ve taken so we can reproduce what you’re seeing in a 1:1 environment please! The more detail/steps shared will help us produce what you’re seeing faster which in return will help us find a solution faster as well.
We’ll get to the bottom of this !!
Valmir