vmware-iso: SSH timeout VMWare Workstation 11 (DHCP no leases)

[Darren Hartford]

Hi All,

Using packer 0.7.5 on Centos 6.5 with the vmware-iso builder (with VMWare Workstation 11).

I can't seem to get around the SSH timeout.  When using the PACKER_LOG=1, I see these in the logs:

2015/05/01 15:00:11 packer-builder-vmware-iso: 2015/05/01 15:00:11 Lookup up IP information...

2015/05/01 15:00:11 packer-builder-vmware-iso: 2015/05/01 15:00:11 DHCP leases path: /etc/vmware/vmnet8/dhcpd/dhcpd.leases

2015/05/01 15:00:11 packer-builder-vmware-iso: 2015/05/01 15:00:11 IP lookup failed: IP not found for MAC in DHCP leases

2015/05/01 15:00:11 packer-builder-vmware-iso: 2015/05/01 15:00:11 Error getting SSH address: IP lookup failed: IP not found for MAC in DHCP lease

Repeated multiple times.  

Prior attempts, tried with Fedora20/VMPlayer 6 with headless=false, and it works when I'm running on the box with the UI (not desired, looking for headless, but proves it worked).  With VMPlayer 6 headless=true, getting SSH timeouts.  When I try a fresh build of CentOS 6.5, Packer 0.7.5, VMWare Workstation 11 (all fresh, only thing on the box), still getting the SSH timeout.

Please let me know anything I should try.

-D

[Alvaro Miranda Aguilera]

Hello

I use packer with virtualbox over redhat 7 and have no issues

can you send a zip with a packer project? I will like to have a look
and can give it a run

> --
> You received this message because you are subscribed to the Google Groups
> "Packer" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to packer-tool...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Darren Hartford]

I did some more digging (VNC into the in-progress while waiting for SSH timeout).  The SSH timeout is because it never handled the install, hanging on the kickstart file download for CentOS.

When I do a netstat -a, it appears the HTTP host that packer starts is starting only on TCP6 (IPv6) addresses, and not an IPv4 address (which the .HTTPIP is an IPv4 address...assuming it would be the correct address (vmnet8 host ip address).

Question: How can I force the packer HTTP_DIRECTORY web server to start and host through IPv4 address (preferably the one that the .HTTPIP argument is expecting)?

-D

[Darren Hartford]

SSH timeout related to HTTP_Directory only hosting on IPv6, but not IPv4, the .HTTPIP parameter is an IPv4 address.

Tried on Centos, but still packer is hosting the HTTP_Directory on IPv6 address (i.e. netstat -a |grep <port expected for http_directory>  ):

sysctl -w net.ipv6.conf.all.disable_ipv6=1
sysctl -w net.ipv6.conf.default.disable_ipv6=1

It appears Docker has similar challenges: https://github.com/docker/docker/issues/2174

-D

[Ethan S.]

I've also run into this fun bug ONLY on headless installs. It seems under virtualbox or VMware workstation you may also need to DISABLE 3d acceleration (something like vm.vmw["mks.3denable"=false]) otherwise the system tries to load the 3D driver, only to crash when there isn't any GUI to utilize it. Its a weird bug, but I've been able to consistently reproduce it.

[Joe Arnet]

Try changing the nic setting in vmware workstation - it needs to be nat. 

On Friday, May 1, 2015 at 5:34:15 PM UTC-4, Darren Hartford wrote: