All committers have signed the CLA.

Coverage increased (+9.2%) to 24.254% when pulling 718a65d on mshaverdo:extended_configmap into f3b9e29 on aledbf:master.

Can you mark this as fixes #12 and fixes #77?

I'm not a fan of the new format, but it is at least backwards compatible. If we break backwards compatibility we can redesign it as a list instead of map. Let's see what @aledbf thinks.

@mshaverdo @steven-sheehy I don't like the new format. Right now, using a configmap makes no sense with all the features we are trying to define in a key-value string definition. I think is time to evolve using a no namespaced CRD

@mshaverdo @steven-sheehy Maybe, it's a good idea to summarize desired features?

• multiple services on the same IP (Add support for multiple services on the same ip #12 How to configure: multiple services on the same ip address #55)
• multi-nic systems (Add support for multi-nic systems #13 Ability to use specific interface for each VIP #77)
• ...anything else?

Maybe IPv6 support #64?

Looks like multiple nics will also fix #73.

@mshaverdo @steven-sheehy I don't like the new format. Right now, using a configmap makes no sense with all the features we are trying to define in a key-value string definition. I think is time to evolve using a no namespaced CRD

👍 concur on CRD
I met the same situation. external IP is quite valuable, reuse the same external IP in NAT mode, map different backend service to different host port is a tradeoff.

another major comment for this implementation:
image there's two lines in the configMap, one says using "interface:eth0", another says "interface:eth1".

But interface is a property of vrrp_instance . There's single vrrp_instance(named vips ) generated in the /etc/keepalived/keepalived.conf. For different items in the configMap, each item will have their own virtual_server, but belongs to the same vrrp_instance .
So in above case, I believe the eth1 will override the eth0 line in configMap.( I did not try )

so I think the resolution for diff NIC interface is : a global config item in CRD...

another major comment for this implementation:
image there's two lines in the configMap, one says using "interface:eth0", another says "interface:eth1".

But interface is a property of vrrp_instance . There's single vrrp_instance(named vips ) generated in the /etc/keepalived/keepalived.conf. For different items in the configMap, each item will have their own virtual_server, but belongs to the same vrrp_instance .
So in above case, I believe the eth1 will override the eth0 line in configMap.( I did not try )

so I think the resolution for diff NIC interface is : a global config item in CRD...

forgive my ignorance. iface can be VIP granularity .

  virtual_ipaddress {
    1.2.3.4 dev ens192
    1.2.3.5 dev ens224
  }