[caasp-beta] beta3 try1 - cluster bootstrap stuck

Alejandro Bonilla abonilla at suse.com
Fri May 12 20:01:34 MDT 2017


Hi Denis,

> On May 12, 2017, at 6:25 PM, Denis Kulagin <deniskulagin at foxwilder.com> wrote:
> 
> Hi All,
> 
> Just found out about this list. And wanted to share my experience here and ask a few questions.
> 
> Trying beta3 now. First downloaded VMware image, booted it and... nothing happens :) Only port 22 is open, couldn't find root password. Pinged my friend David, who recommended installing from ISOs. Though I'm lazy - I actually did. Installation went smoothly on blank VMware image. However, you have to say I will install OS later, do not use easy/auto install VMware offers, it messes up the installer.

The variant images are provided for an automated installation which requires the use of cloud-init which would inject the information at boot. There is some documentation available in one of the links at the Beta site, but this is still WIP.

> 
> I was able to log into controller node. Created account. 
> 
> Created a brand new VMware image with the same ISO for the first worker node. Installed with "autoyast=http://<my_ip>/autoyast netsetup=dhcp". It did show up in the console with the hostname linux. At this point I decided to create one more to have dedicated master. Did the same exact install for the second worker, it finished successfully, but never showed up in the console.

I can see two reasons but I’ll have to assume things:
- You created a template, which carries the system id and then would be seen as a duplicate.
- A hostname was not provided for the VM which makes the minion to use the ‘linux’ hostname by default. A hostname was not provided by DNS either.

I’d recommend to boot specifying the hostname at install/boot time or configuring it the yast2 network section while at the installer.

> 
> So now I have only one node with id 1 and hostname 'linux' and I try to bootstrap. Its been hanging out for an hour now and I believe its stuck.

A master and a worker is mandatory. It will stay there… I’d suggest to reinstall the admin and then create 2 new VMs which are booted/installed via the ISO pointing to the controller IP or using the autoyast path as you did before.

> 
> One more thing to mention - I did not provide NTP server.

In a “lab” this may not be necessary if both VMs or on the same host. They should be given the same time as the hw clock. If this is on an ESX cluster, then you certainly need to specify one when configuring the initial admin node.

> 
> Any ideas? I'm thinking I will try this scenario one more time tomorrow with the ntp server. And then with just one worker node (somehow I'm suspicion about both working may have the same 'linux' hostname…).

I think you are onto something ;-)

> 
> 
> Denis Kulagin
> deniskulagin at foxwilder.com
> +1 (212) 706-0172
> 
> 
> _______________________________________________
> caasp-beta mailing list
> caasp-beta at lists.suse.com
> http://lists.suse.com/mailman/listinfo/caasp-beta



More information about the caasp-beta mailing list