maas-devel team mailing list archive

[Dafang Zhang <me2050@xxxxxxxxx>]

Thanks to Julian for pointing out that this issue could be due to the bug (linked in his reply below).  I checked the provision-agent.log on the zookeeper node and found that the exceptions matched the bug description.  So I followed the workaround and it did work.  Sounds like the latest juju packages fixed the original issue I had.  Here're the steps that worked for me.  Hope this help those who would have similar issues.  

(All nodes are VirtualBox VM's)
1. juju destroy-service <any previous services>

2. juju destroy-environment

3. juju terminate-machine <all previous Ready-state nodes, including the zookeeper node>
4. <modify the environments.yaml to set default-series: quantal>
5. <modify /etc/maas/import_pxe_files to set RELEASES="quantal">
6. sudo maas-import-pxe-files
7. <boot nodes>

8. <from maas web ui, the nodes appear as Declared>
9.<from maas web ui, click the setting button > select 12.10 for both Commissioning and Deploy > Save >
10.<from maas web ui, edit each node to select 12.10 under Release > Save >
11.<from maas web ui, accept and commission each node; the nodes are transitioned to Commissioning state>
12.<boot each node once; wait for the nodes to enlist and shutdown; from maas web ui, verify that the nodes are transitioned to Ready state;>
13. juju bootstrap
14.<from maas web ui, verify that one of the Ready-state node is transitioned to Allocated state>
15.<boot that node; it PXE-boot, complete the server install and becomes the zookeeper node>
16.<juju status should now be able to connect to the environment and display the zookeeper node as machine 0>
17.<modify the environments.yaml to set default-series: precise>
17.juju deploy <services>

18.<from maas web ui, verify that the rest of Ready-state nodes are transitioned to Allocated to <user> state immediately after the juju deploy command lines.>
19.<boot the nodes another time while they are in Allocated state, now they are able to PXE-boot, complete the server install and juju service deployment>
20.juju status should now have similar output as the following:
ubuntu@ubuntu:~$ juju status

2012-11-01 00:12:17,803 INFO Connecting to environment...
2012-11-01 00:12:18,857 INFO Connected to environment.
machines:
  0:
    agent-state: running
    dns-name: 192-168-77-20.master
    instance-id: /MAAS/api/1.0/nodes/node-7dccdfc4-2387-11e2-833f-080027edacc9/
    instance-state: unknown
  1:
    agent-state: running
    dns-name: 192-168-77-21.master
    instance-id: /MAAS/api/1.0/nodes/node-cae5ea18-238b-11e2-a832-080027edacc9/
    instance-state: unknown
  4:
    agent-state: running
    dns-name: 192-168-77-22.master
    instance-id: /MAAS/api/1.0/nodes/node-4eb5a1e6-238f-11e2-a832-080027edacc9/
    instance-state: unknown
services:
  mysql:
    charm: cs:precise/mysql-9
    relations: {}
    units:
      mysql/0:
        agent-state: started
        machine: 1
        public-address: 192-168-77-21.master
  tomcat7:
    charm: cs:precise/tomcat7-1
    exposed: true
    relations:
      cluster:
      - tomcat7
    units:
      tomcat7/0:
        agent-state: started
        machine: 4
        open-ports:
        - 8080/tcp
        public-address: 192-168-77-22.master
2012-11-01 00:12:19,161 INFO 'status' command finished successfully



----- Original Message -----
From: Julian Edwards <julian.edwards@xxxxxxxxxxxxx>
To: Dafang Zhang <me2050@xxxxxxxxx>
Cc: 
Sent: Wednesday, October 31, 2012 11:00 AM
Subject: Re: [Maas-devel] Ready State not Changed to Allocated after Juju Deploy

On Wednesday 31 Oct 2012 07:30:09 Dafang Zhang wrote:
> [Sorry again, missed the subject line for the new thread.  Added.]
> 
> > ----- Original Message -----
> > From: Julian Edwards <julian.edwards@xxxxxxxxxxxxx>
> > To: Dafang Zhang <me2050@xxxxxxxxx>
> > Cc: 
> > Sent: Wednesday, October 31, 2012 4:47 AM
> > Subject: Re: [Maas-devel] Ready State not Changed to Allocated after Juju
> > Deploy
> > 
> > 
> > [Please start new threads with a new email, don't reply to existing emails
> > by  changing the subject as threaded email clients still see it as part
> > of the old thread.]
> 
> [Thanks Julian for pointing it out.  Sorry about that.  Starting a new
> thread with new conversations below..]
> > On Tuesday 30 Oct 2012 21:01:21 Dafang Zhang wrote:
> >> Hi,
> >>
> >> 
> >>
> >> I think I'm pretty close to a successful run with the quantal-proposed
> >> package (bzr1269).  Here's the issue - I got the zookeeper node (machine
> >> 0)
> >> up and running, and it showed up as Allocated to <user> from MAAS web ui.
> >>  I have another node in Ready state.  Then, after "juju deploy mysql",
> >> "juju status" shows the machine status as pending.  The node state from
> >> the
> >> MAAS web ui is still in Ready state.  I booted the node, but it stuck at
> >> the PXE booting, "Trying to load: pxelinux.cfg/<xx-xx-xx-xx-xx-xx>". 
> >> After
> >> a while, it went through the enlisting PXE boot again, then the node
> >> shuts
> >> down.  I noticed there're a few defects have been committed for the the
> >> next 12.10-stabilization release, but wondering if anyone could shed me
> >> some lights on how to resolve this issue in the current quantal-proposed
> >> build?  
> >>
> >> 
> >>
> >> Thanks in advance for any insight!
> >>
> >> 
> >>
> >> Dafang
> > 
> > Sounds like you configured your environments.yaml incorrectly; did you
> > put 
> > "localhost" as the maas server address?  If so, put the real host name or
> > IP  address.
> 
> In my environments.yaml, I did use the MAAS server address, which is the
> same as where it's set for the maas-region-controller and
> maas-cluster-controller.  Here's the actual content of the file:
> 
> default: maas
> environments:
>   sample:
>     type: ec2
>     control-bucket: juju-53a879aae42946b2ae9a62ed68670ae7
>     admin-secret: 8bbaf87f595c481c860f7855f3566a8e
>     default-series: precise
>     ssl-hostname-verification: true
>   maas:
>     type: maas
>     control-bucket: juju-53a879aae42946b2ae9a62ed68670ae7
>     maas-server: 'http://192.168.77.101/MAAS'
>     maas-oauth:
> 'wWB47CUWbBuXKjVdKa:Dk4hYY38uXABTsz49T:DZajeZLczsrjhSYyndu6BMKzkrQqyNsT'
> admin-secret: 'nothing'
>     default-series: precise
>     juju-origin: ppa

This might be bug https://bugs.launchpad.net/maas/+bug/1068086

Can you check the provisioning agent log on the zookeeper node to see if it 
matches that one please?

J