At https://flynn.io/docs/installation/manual:
If you are starting more than one node, the cluster should be configured using a discovery token. flynn-host init is a tool that handles generating and configuring the token.
Does this mean that I can actually skip sudo flynn-host init
commands if I'm going single-node?
Later on, when I run flynn-host bootstrap
, should I omit --discovery-url
if going single-node?
If I run discovery commands and pass --discovery-url
, I get:
14:56:17.184567 check online-hosts
14:58:18.095882 check online-hosts error: timed out waiting for 1 hosts to come online (currently 0 online)
The following hosts were discovered but remained unreachable:
And if I omit --discovery-url
, I get:
15:00:31.207176 check online-hosts
15:02:31.253575 check online-hosts error: timed out waiting for 1 hosts to come online (currently 0 online)
The following hosts were discovered but remained unreachable:
http://127.0.0.1:1113
I got flynn-host
running, but it doesn't listen to that port (it doesn't seem to listen any port)... should it?
Trying to flynn create example
gives:
15:23:54.433097 main.go:172: no clusters configured
My instance is in intranet, configured into an internal DNS server.
I have (now) http_proxy
, https_proxy
, HTTP_PROXY
and HTTPS_PROXY
environment variables set (to survive through the installation), but where/how should I configure them for flynn?
It is probably a bug (#3640) in the current version of flynn, maybe just try v20160918.1
.
Does this mean that I can actually skip
sudo flynn-host init
commands if I'm going single-node?
Yes.
Later on, when I run
flynn-host bootstrap
, should I omit--discovery-url
if going single-node?
Yes.
From your issue it sounds like the flynn-host process is not running or there is a restrictive firewall configured. Also, Flynn does not currently support proxies (see #1973).
Alright, fair enough. I guess I'll wait until proxy support is available...
I admit I was confused as well. Is it still necessary to create a CNAME DNS record with one node? The only thing that I see the DNS setting being used is for service discovery
@webmutation To access an application running in Flynn you need a working route to it.
In order to resolve applications in flynn dynamically, you require a wildcard DNS entry:
i.e. <appname>.my.flynn.cluster.com
, so you need DNS settings in place which resolve *.my.flynn.cluster.com
to your flynn cluster nodes.
If you have only one node, you would have a DNS entry of *.my.flynn.cluster.com
pointing to the IP of your single node.
Most helpful comment
@webmutation To access an application running in Flynn you need a working route to it.
In order to resolve applications in flynn dynamically, you require a wildcard DNS entry:
i.e.
<appname>.my.flynn.cluster.com
, so you need DNS settings in place which resolve*.my.flynn.cluster.com
to your flynn cluster nodes.If you have only one node, you would have a DNS entry of
*.my.flynn.cluster.com
pointing to the IP of your single node.