We are facing RPC issues related to our multiregional tyk setup.
Our multi-cloud tyk gateway setup:
Because we have separate redis instances after doing a change in a dashboard API definition, only 1 tyk gateway are able to detect and apply changes: level=warning msg=“[RPC STORE] Received Reload instruction!”
Other instance is keeping showing messages until service will be restarted manually: level=info msg=“RPC is alive”
Is it possible to use same setup but to avoid having these issue?
There are plans to have 10+ regions and it will be quite convenient to use separate redis per region.
It looks like both gateways regardless of region might have the same group_id. This could be why they are both not receiving updates - because Tyk’s management layer see them as the same cluster.
Try setting a different group id to each cluster of gateways / region in your tyk.conf or via environment variable.