Micro bosh manifest changes when upgrading from stemcell 0.8.1. to 1.5pre1376

28 views
Skip to first unread message

Clete Penrod

unread,
Nov 21, 2013, 2:29:28 PM11/21/13
to bosh-...@cloudfoundry.org
Running into trouble when trying to deploy a micro bosh instance from the bosh-stemcell-1376-vsphere-esxi-ubuntu.tgz stemcell to vsphere.

When I use bosh 1.0.3 micro setup, i can deploy the 0.8.1 stemcell using the attached manifest with no problems. When I switch to bosh 1.5.0.pre.1381 using the updated micro plugin and attempt to deploy the bosh-stemcell-1376-vsphere-esxi-ubuntu.tgz, I get the following error:


 Creating VM from...                 |oooo                    | 2/11 00:01:10  ETA: 00:03:25/home/vmware/.rvm/gems/ruby-1.9.3-p327/gems/bosh_vsphere_cpi-1.5.0.pre.1381/lib/cloud/vsphere/cloud.rb:980:in `create_nic_config_spec': Can't find network: DIZ 192.168.192.0/24 (RuntimeError)

Are there changes in the new stemcell with how you define the network name in the manifest?
bosh_micro.yml

Clete Penrod

unread,
Nov 21, 2013, 2:36:20 PM11/21/13
to bosh-...@cloudfoundry.org
Additional info:

gem list bosh:


bosh-registry (1.5.0.pre.1381, 1.5.0.pre.1376)
bosh-stemcell (1.5.0.pre.1381, 1.5.0.pre.1376)
bosh_aws_cpi (1.5.0.pre.1381, 1.5.0.pre.1376)
bosh_cli (1.5.0.pre.1381, 1.5.0.pre.1376)
bosh_cli_plugin_micro (1.5.0.pre.1381, 1.5.0.pre.1376)
bosh_common (1.5.0.pre.1381, 1.5.0.pre.1376)
bosh_cpi (1.5.0.pre.1381, 1.5.0.pre.1376)
bosh_openstack_cpi (1.5.0.pre.1381, 1.5.0.pre.1376)
bosh_vcloud_cpi (0.4.9)
bosh_vsphere_cpi (1.5.0.pre.1381, 1.5.0.pre.1376)



Troy Astle

unread,
Nov 22, 2013, 5:52:12 AM11/22/13
to bosh-...@cloudfoundry.org
This is what I am using to define my network in micro_bosh.yml using VSphere and stemcell 1376:

network:
  type: manual
  ip: 9.10.4.11
  netmask: 255.255.255.0
  gateway: 9.10.4.1
  dns:
  - 144.254.71.184
  cloud_properties:
    name: CPG-MGMT # Network name in VSphere

Only real difference I can see is that you are passing the subnet after the VSphere network name.

Troy.

William C Penrod

unread,
Nov 22, 2013, 4:51:02 PM11/22/13
to bosh-...@cloudfoundry.org
Thanks for the working manifest to review.


I was snooping the deployment soap log and ran across this tidbit. 

This is the soap request call for the cloud properties network name. The one that is currently working for me has the slash character encoded.

<soapenv:Body><FindByInventoryPath xmlns="urn:vim25"><_this type="SearchIndex">SearchIndex</_this><inventoryPath>DC1/network/DIZ 192.168.192.0%2f24</inventoryPath></FindByInventoryPath></soapenv:Body>

The request on the deploy that does not work is not encoded.

<soapenv:Body><FindByInventoryPath xmlns="urn:vim25"><_this type="SearchIndex">SearchIndex</_this><inventoryPath>DC1/network/DIZ 192.168.192.0/24</inventoryPath></FindByInventoryPath></soapenv:Body>

I don't think vsphere is finding the network name because its interpreting the forward slash as part of path, not the name.



To unsubscribe from this group and stop receiving emails from it, send an email to bosh-users+...@cloudfoundry.org.

William C Penrod

unread,
Nov 22, 2013, 6:09:24 PM11/22/13
to bosh-...@cloudfoundry.org
The work around for this is to just add the character encoding to the network name in the manifest file.

cloud_properties:
    name: DIZ 192.168.192.0%2f24

Reply all
Reply to author
Forward
0 new messages