Guys, 

Not sure if anyone has done some more testing with CoreOS. I am currently doing 
a PoC and having some issues with building the cluster. 

The template itself is working okay and the user supplied ssh keys are properly 
registered with the instance. However, the etcd cluster is not forming 
properly. The instances can see each other and there is no fw to block the 
requests. 

I am using addr and peer-addr settings in the config-cloud.xml in the following 
manner: the first instance will have addr and peer-addr to the same IP - 
10.0.1.x:4001 and 10.0.1.x:7001. The remaining instances will have the same 
addr value, but the peer-addr reflects the instance's internal ip address. 

Upon starting the instances, i can see that all hosts join the cluster - 
journalctl -u etcd -f tell me that the instances join the cluster. However, I 
am unable to do anything with the cluster. For instance, when I try to query 
the machines, it times out: 

etcdctl --debug=true ls /_etcd/machines --recursive 
Cluster-Peers: http://10.0.1.45:4001 http://10.0.1.45:4001 
http://10.0.1.45:4001 http://10.0.1.45:4001 
Curl-Example: curl -X GET 
http://10.0.1.45:4001/v2/keys/_etcd/machines?consistent=true&recursive=true&sorted=false
 

Perhaps someone can tell me what I am missing? 

Andrei 

----- Original Message -----

> From: "Phillip Kent" <phillip.k...@xmlsoup.com>
> To: "<users@cloudstack.apache.org>" <users@cloudstack.apache.org>
> Sent: Thursday, 5 February, 2015 11:07:14 PM
> Subject: Re: Docker Support?

> Thanks all for this useful thread.

> I can confirm that a CoreOS template runs fine on an ESXi hypervisor
> in CloudStack 4.3.1. The type is set to "Other (64-bit)". I didn't
> set
> up the template so I don't know about relative performance with other
> type settings.

> Regarding CoreOS, does anyone know what port(s) are required for
> CoreOS to do auto-updating?

> Obviously I can set an egress rule to allow all ports but I might
> want
> to be more restrictive.

> -Phillip

Reply via email to