We should document the three types of traffic from outside to the inside and how people achieve that
- HTTP/TLS -> router (SNI special case)
- TCP -> service nodeport and a ramp node / vrrp setup (basically, use the same nodes as the router but advertise them with VRRP and use the DNS address of the router nodes)
- Connecting pod network directly to the outside (partially documented in the ramp node -> SDN issue)
All three of these are "supported", but they need doc. They are really three facets of the same problem.
@knobunc do you have a card that covers this doc? I think this is one topic, three subsections, and then we can stop answering this question all the time :)
For 2 the recommendation is:
- Setup router nodes (either TCP load balance or ipfailover VRRP setup) in an HA config
- When asking for nodeports, use the router nodes with their VIPs (VRRP) or with a range TCP load balancer (cloud, F5) as the destination IP. You can also use the wildcard DNS for the router + the nodeport
We should document the three types of traffic from outside to the inside and how people achieve that
All three of these are "supported", but they need doc. They are really three facets of the same problem.
@knobunc do you have a card that covers this doc? I think this is one topic, three subsections, and then we can stop answering this question all the time :)
For 2 the recommendation is: