> On Apr 21, 2015, at 3:19 PM, cs user <[email protected]> wrote: > > Hi Phillip, > > Many thanks for getting back to me. I managed to get it working by > following this guide yesterday (might be useful for someone else): > > http://help.ninefold.com/servers/how_to_create_a_coreos_server/ > > I put the user-data into a file with my generated token and then convert it > to base64 with: > > base64 -w 0 coreos-userdata > > Server was then launched with cloudmonkey: > > deploy virtualmachine -userdata=base64stringFromAbove -name core01 -(other > params for zone, template etc) > deploy virtualmachine -userdata=base64stringFromAbove -name core02 -(other > params for zone, template etc) > > etc...... > > Just need to script this now to automatically create farms. Seems to work > fine! > > One piece of good advice seems to be that if you are setting up a CoreOS >> cluster using fleet, if any of the VMs fail to deploy completely, the best >> thing is to get a fresh discovery token and re-deploy all of the VMs. It >> does not seem to be easy to clear a 'failed' record from the discovery >> database. > > > I found this as well, great to hear it wasn't just me. >
just want to point out that this is a coreOS discovery mechanism “issue”, cloudstack is involved there. > Cheers! > > > > > On Tue, Apr 21, 2015 at 1:35 AM, Phillip Kent <[email protected]> > wrote: > >> Hi cs user, >> >> I have not so far had any issues with using discovery and passing the >> discovery URL via userdata. >> >> If you are using a 'Private with Gateway Services' network this simple >> 'cloud-config' file should work: >> >> https://github.com/Interoute/CoreOS-and-VDC/blob/master/cloud-config-template-simple >> >> >> And the new VM will pull its config from the virtual router. My CS setup >> uses Advanced Networking, so I don't know how this goes with the Basic >> networking/Security groups. >> >> One piece of good advice seems to be that if you are setting up a CoreOS >> cluster using fleet, if any of the VMs fail to deploy completely, the best >> thing is to get a fresh discovery token and re-deploy all of the VMs. It >> does not seem to be easy to clear a 'failed' record from the discovery >> database. >> >> BTW, I will be publishing soon a couple more blogs on CoreOS clusters and >> fleet. Keep an eye on: >> http://cloudstore.interoute.com/main/knowledge-centre/blog >> >> - Phillip >> >> ---------- Forwarded message ---------- >> From: cs user <[email protected]> >> To: [email protected] >> Cc: >> Date: Mon, 20 Apr 2015 10:07:54 +0100 >> Subject: >> Ah, I managed to get the sshkeypair working with this template! :) >> >> http://dl.openvm.eu/cloudstack/coreos/x86_64/ >> >> Just need to get discovery working now. >> >> >> On Mon, Apr 20, 2015 at 8:58 AM, cs user <[email protected]> wrote: >> >>> Hi All, >>> >>> Thanks for the blog post Phillip, very impressive! :) >>> >>> I've had a play with coreos/cloudstack myself, but it's been pretty >> manual >>> and the SSH public key has been baked into the template. >>> >>> I guess I'm missing something very simple here, but how is everyone >>> managing to set the following via the dhcp router metadata? >>> >>> ssh_authorized_keys: >>> - ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQC0g+ZTxC7weoIJLUafOgrm+h. >>> >>> >>> Are you creating a script on the coreos template which curls against the >>> router the first time it boots? >>> >>> Cheers! >>> >>> >>
