Closed pankaj-dahiya-devops closed 2 years ago
@dgrove-oss Do you have any recommended solution for this!
The assumption is that there is a stable DNS hostname (or IP address) to use for ingress for the lifetime of the deployment.
But this assumption has a very critical exception of restarting the machine which gave it a different IP address. So how to bypass this situation!
This needs to be fixed outside of openwhisk. You need to either (a) deploy in an environment with proper DHCP configuration that includes DNS resolution and use the hostname as the whisk.ingress.apiHostName
or (b) you need to use a static ip addresses (so the host machine gets the same IP address across reboots).
Where are you deploying openwhisk @pankaj-dahiya? Typically, as Dave noted, there is a stable domain name and if the IP address changes, one updates the DNS record for the deployment. If you don't have a static IP, how do you communicate the change of IP to the users of your openwhisk cluster?
Hi @rabbah, I am deploying this in an isolated environment, where users will access the app via private IP (that will be static). But there could be a reason, that in future the IT team may consider a change in IP address (another static IP). In that case, all other services will still work fine, but not open whisk, and thus they must have to manually handle this thing! That's why I opened this ticket, which is a kind of feature request.
Dear team,
I knew there is a small advantage we get by asking the user to provide whisk.ingress.apiHostName at day zero... But this is actually not feasible. Because once user deploys the helm chart by providing whisk.ingress.apiHostName and in future, if machine restarts its IP may be changed. And if that happens then our cluster which will automatically start running after machine restart will fail due to IP mismatch.
So want to know what open whisk team or other guys are handling this situation.