Hoping there are enough Kubernetes folks following me that one of you can help...
I'm trying to install the Kubernetes dashboard on an OKE cluster. There's a docs page: https://docs.oracle.com/en-us/iaas/Content/ContEng/Tasks/contengstartingk8sdashboard.htm - it mentions that the Helm chart creates a new namespace for the dashboard, I initially thought maybe that was an issue, so ran the helm chart with the kube-system namespace. Same result both times (details next toot).
=> More informations about this toot | More toots from dneary@mastodon.ie
OK - steps I follow:
=> More informations about this toot | More toots from dneary@mastodon.ie
kubectl logs gives this error - I have no idea how to fix this. Is this a config issue?
nginx: the configuration file /kong_prefix/nginx.conf syntax is ok
nginx: [emerg] bind() to [::1]:8444 failed (99: Cannot assign requested address)
nginx: configuration file /kong_prefix/nginx.conf test failed
=> More informations about this toot | More toots from dneary@mastodon.ie
@dneary Any chance IPv6 is disabled on the machines?
=> More informations about this toot | More toots from kevin@km6g.us
@kevin Shouldn't be... How would I know?
=> More informations about this toot | More toots from dneary@mastodon.ie
@dneary Simplest way is to replace [::1] in that configuration file with 127.0.0.1. If the result works, then IPv6 is disabled somehow.
=> More informations about this toot | More toots from kevin@km6g.us
@kevin the Kong configuration file? It's pulled in with the dashboard Helm chart - how do I change it?
Could it be an arm64 vs amd64 thing?
=> More informations about this toot | More toots from dneary@mastodon.ie
@dneary No, this is not related to the architecture.
Either something is already listening on port 8443, or IPv6 'localhost' ([::1]) is not available. Those are the two reasons that NGINX would report an inability to bind to that address.
=> More informations about this toot | More toots from kevin@km6g.us
@kevin Interesting... And would this be in the host VM, or in the container? Overlay network cofiguration or the container image, I suppose...
=> More informations about this toot | More toots from dneary@mastodon.ie
@kevin Also: I found this StackOverflow post saying a setting might be needed in /etc/sysctl.conf:
[#] allow processes to bind to the non-local address
[#] (necessary for apache/nginx in Amazon EC2)
net.ipv4.ip_nonlocal_bind = 1
(presumably there is an equivalent ipv6 setting too). Issue is, these are not my containers, I'm not sure how to make config changes to Kong.
=> More informations about this toot | More toots from dneary@mastodon.ie
@kevin Unsurprisingly, it's a known issue. Easy fix, hopefully (will try later): https://github.com/kubernetes/dashboard/issues/8765
=> More informations about this toot | More toots from dneary@mastodon.ie
@dneary Ahh, port conflict.
=> More informations about this toot | More toots from kevin@km6g.us
@kevin Victory is mine!
=> More informations about this toot | More toots from dneary@mastodon.ie This content has been proxied by September (ba2dc).Proxy Information
text/gemini