[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Install and add a node



On Nov 6, 2013, at 16:13 , Juanje Ojeda Croissier <juanje ojeda gmail com> wrote:

> I tried a fresh broker's installation with the oo-install and it ask me for a SSH key or password for a local instalation:
> http://paste.fedoraproject.org/52204/77223613/
> 
> Before it didn't ask me for it. Should it be that way?

It should be fixed now. The 'hostname' now asks for an FQDN, but "SSH host" can still be localhost, and the installer will assume that you are doing a local installation in that case.

> 
> 
> 
> On Wed, Nov 6, 2013 at 8:22 PM, Arunabha Ghosh <arunabha gh gmail com> wrote:
> Hi Harrison, 
>                   The output is here  http://ur1.ca/fzspg
> 
> The weird thing is, the ssh succeeds yet the installer thinks it has failed.
> 
> D, [2013-11-06T12:19:03.274599 #16435] DEBUG -- net.ssh.authentication.methods.publickey[3fc1c05a4da0]: publickey succeeded (a6:1a:d3:8a:e1:61:14:11:62:2d:17:e4:c3:d7:0a:ed)
> D, [2013-11-06T12:19:03.274859 #16435] DEBUG -- tcpsocket[3fc1c0415f34]: queueing packet nr 8 type 90 len 44
> 
> Could not connect to broker.paas.arunabha.com with user fedora. You
> must set up an SSH key pair and using ssh-agent is strongly
> recommended.
> 
> 
> Thanks,
> Arunabha
> 
> 
> On Wed, Nov 6, 2013 at 6:43 AM, N. Harrison Ripps <hripps redhat com> wrote:
> Hey Arunabha--
> 
> On Nov 5, 2013, at 20:28 , Arunabha Ghosh <arunabha gh gmail com> wrote:
> 
> > Ok, Thanks Harrison. Will wait for the debug enabled installer.
> 
> The installer now has a '-d' flag that will enable debugging output. This is not a full implementation; right now it just turns on the fire hose for SSH connections. Hopefully this will help us figure out why you are running into problems.
> 
> As a reminder, if you want to pass multiple flags to the oo-install curl invocation, you need to enclose all of the flags in a single quoted string:
> 
> sh <(curl -s http://oo-install.rhcloud.com/) "-d -c my-alternate-config-file.yml"
> 
> Give this a shot and pastebin the output.
> 
> Thanks,
> Harrison
> 
> > On Tue, Nov 5, 2013 at 5:03 PM, Harrison Ripps <hripps redhat com> wrote:
> >
> >
> > On Nov 5, 2013, at 19:39, Arunabha Ghosh <arunabha gh gmail com> wrote:
> >
> >> Trying it from my local system to deploy a fresh install fails with
> >>
> >> [substituting example.com for my actual domain]
> >>
> >> Validating fedora broker example com   
> >> Could not connect to broker.example.com with user fedora. You
> >> must set up an SSH key pair and using ssh-agent is strongly
> >> recommended.
> >>
> >> I have set up the key pairs appropriately and have a .ssh/config file entry for the broker which has the appropriate Identityfile. I can ssh to the host without a password using
> >>
> >> ssh fedora broker example com
> >>
> >>
> >> Why does the installer fail to connect ?
> >
> > That is a good question. The installer uses a combination of Net::SSH and direct system calls to ssh, so I'm a little stymied. Tomorrow I can provide you with an installer package that has debugging cranked up and hopefully we can figure out what the issue is.
> >
> >>
> >> On Tue, Nov 5, 2013 at 12:25 PM, N. Harrison Ripps <hripps redhat com> wrote:
> >> The Add a Node workflow for Origin should be in better shape now. Give it a shot and let me know if yo run into any issues.
> >>
> >> On Nov 5, 2013, at 10:03 , Juanje Ojeda Croissier <juanje ojeda gmail com> wrote:
> >>
> >> > On Tue, Nov 5, 2013 at 2:54 PM, N. Harrison Ripps <hripps redhat com> wrote:
> >> > Hey Juanje--
> >> >
> >> > On Nov 5, 2013, at 9:40 , Juanje Ojeda Croissier <juanje ojeda gmail com> wrote:
> >> >
> >> > > Hi Harrison, thanks for the help here :-)
> >> > >
> >> > > On Tue, Nov 5, 2013 at 12:42 AM, Harrison Ripps <hripps redhat com> wrote:
> >> > > Hey Juanje--
> >> > >
> >> > > On Nov 4, 2013, at 19:13, Juanje Ojeda Croissier <juanje ojeda gmail com> wrote:
> >> > >
> >> > >> Hi
> >> > >>
> >> > >> I'm sorry if this is a known and silly question but I'm a bit lost.
> >> > >>
> >> > >> I've created a broker with the oo-install and now I want to create a node to be used by the broker.
> >> > >> Should I use the oo-install in the node but removing the role broker? Should I run the oo-install in the broker telling it which IP is the one from my new node?
> >> > >> Is there another way?
> >> > >
> >> > > When you set up the broker, did you do it as a localhost installation? If so, run the Add a Node routine straight on the new Node host and provide info about the existing broker instance plus the Node settings from there.
> >> > >
> >> > >
> >> > > I did the localhost instalation (in a DigitalOcean droplet) for the Broker. Now I tried to oo-install a new node in a new droplet but not luck.
> >> > > After a few tries (sorry but I found the questions a bit confusing) I got the configuration ready but I got this error:
> >> > >
> >> > > Deploying workflow 'origin_add_node'.
> >> > > Setting up htpasswd for default user account.
> >> > > The list of nodes in the config file at /root/.openshift/oo-install-cfg.yml does not contain an entry for 0. Exiting.
> >> > >
> >> > > This is the config file the installer has generated:
> >> > > https://paste.fedoraproject.org/51734/66154013/
> >> >
> >> > Okay, that looks plausible, but seeing this config makes me realize that using 'localhost' as the host info will only really work for an all-in-one system. If you ever want it to be part of a larger system, we probably have to enforce the use of FQDNs. I'll confirm this with some of the other folks working on the installer right now.
> >> >
> >> >
> >> > Ok, I'll keep this in mind. I was thinking to use a FQDN based hosts in the real installation, but I was checking the install system.
> >> >
> >> > Btw, I did add the hosts (broker and node1) to the /etc/hosts files on each hosts to be able to use the hostnames between them, but it was just a workaround to have this running.
> >> >
> >> > That said, the problem that you are hitting below...
> >> >
> >> > >
> >> > > I had to add a ssh key to be able to connect with the Broker.
> >> > >
> >> > > If you installed the broker from another system (like from your desktop system), then run the Add a Node routine from the same system. This will automatically pick up your broker info from the previous run. Add the new node to the configuration and then the installer will ask you to specify which node host instance you are adding (not as elegant as it could be; I'll work on that!).
> >> > >
> >> > >
> >> > > Well, after try from the Node, I tried from the Broker with similar result:
> >> > >
> >> > > Deploying workflow 'origin_add_node'.
> >> > > Setting up htpasswd for default user account.
> >> > > The list of nodes in the config file at /root/.openshift/oo-install-cfg.yml does not contain an entry for 0. Exiting.
> >> >
> >> > ...this is a straight-up bug; I'll sort this out today.
> >> >
> >> >
> >> > Great :-) Let me know if you need any help or feedback.
> >> >
> >> > --
> >> > Juanje
> >> >
> >> > http://about.me/juanje
> >>
> >>
> >> _______________________________________________
> >> users mailing list
> >> users lists openshift redhat com
> >> http://lists.openshift.redhat.com/openshiftmm/listinfo/users
> >>
> >
> 
> 
> 
> 
> 
> -- 
> Juanje
> 
> http://about.me/juanje



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]