Re: VCL image capture dies, "failed to prepare vmx file"

[Aaron Peeler]

Hello Will,

You will want to disable the auto-assigned mac addresses from ESX/ESXi.

Part of the requirement is to run a dhcp server on the management
node, at a minimum for the private management network.

With that when you add new computers you define your mac addresses
upfront in the add computers interface. When you add multiples you can
add many at a time. Also the add computers tool, will generate
dhcpd.conf entries for each node, so you can cut-n-paste to your
dhcpd.conf file.

Please see the Adding Virtual Machines section in this link for the
detailed steps,
https://cwiki.apache.org/VCL/vcl-221-further-steps-if-using-vmware.html

Hope this helps.
Aaron

On Mon, Jun 25, 2012 at 9:13 AM, William Robinson
 wrote:
> hi all,
>
> new to this list.  i have started setting up vcl here and clemson university
> and this thread has touched on one of the questions i had in trying to do
> so.  do i assign mac addresses based on what esxi generates?  if not, how do
> i map those correctly?  thanks.
>
> will
>
>
>
> On 06/20/2012 10:16 AM, Josh Thompson wrote:
>>
>> -BEGIN PGP SIGNED MESSAGE-
>> Hash: SHA1
>>
>> Michael,
>>
>> Unfortunately, yes, in 2.2.1, you can only set MAC addresses via the add
>> multiple computers page.  You'll need to set them in the database.  This
>> is
>> fixed in 2.3.
>>
>> Josh
>>
>> On Wednesday, June 20, 2012 9:12:52 AM Michael Jinks wrote:
>>>
>>> Thanks Aaron.
>>>
>>> Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
>>> defined for a machine we're capturing *from*.  I figured, well, it's
>>> running and has IP addresses...
>>>
>>> At any rate, I'm looking around the UI and I can't find any page that
>>> will let me enter a MAC address for an existing computer record.  Is
>>> this something I have to do by hand in the database?
>>>
>>> On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:

 Hello Michael,

 One thing that stands out in the log is that your MAC addresses are not
 defined. computer eth0 MAC address: 
 computer eth1 MAC address: 

 This will be needed to correctly create the vmx file and thus get an
 address from the private dhcp server that is should be running on your
 management node.

 Secondly also make sure you have the dhcp server setup on your
 management node for the private network.

 Aaron

 On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks 
>>
>> wrote:
>
> Hi, all.
>
> Trying to resurrect a thread I started on Friday, subject 'Image
> capture
> succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down the
> problem but so far I haven't been able to find a cause or a fix.
>
> I'm trying to capture my first Linux VM. ?Things go along okay through
> the capture stage, but (thanks Mike) we fail when VCL tries to reserve
> and deploy the new image.
>
> The critical message from the log appears to be:
>
> ?|17491|19:19|reload|  WARNING 
> ?|17491|19:19|reload| 2012-06-19
> ?14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx
> file for vcl-linux-template-2-bak on VM host: vcl0001
> ?|17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
> ?|17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
> ?|17491|19:19|reload| (-2) new.pm, process (line: 248)
> ?|17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
> ?|17491|19:19|reload| (-4) vcld, main (line: 346)
>
>
> Further down, I'll paste in a complete copy of the log mail I got from
> this attempt.
>
> Our VM storage settings for this VM host are:
>
> ?Repository Path: ? ? ? /vmfs/volumes/nfs-vcl-templates
> ?Data Store Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> ?VM Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
>
> (Note Data Store and VM are explicitly the same; we've also had VM Path
> set to "(empty)" with no apparent difference in behavior.)
>
> Both of those paths are writable on the hypervisor. ?During capture, a
> VM directory appears in both locations. ?In the directory under the
> Repository path, we have many vmdk files, sizes ranging from 842M to
> 1.5k, with names like:
>
> ?vmwarelinux-linuxtemplate2v1226-v0-s001.vmdk
> ?vmwarelinux-linuxtemplate2v1226-v0-s002.vmdk
> ?vmwarelinux-linuxtemplate2v1226-v0-s003.vmdk
> ?[...]
> ?vmwarelinux-linuxtemplate2v1226-v0-s016.vmdk
> ?vmwarelinux-linuxtemplate2v1226-v0.vmdk
>
> My first guess is that we have one vmdk file for each attempted
> capture,
> and this is a series of xdiffs on the original?
>
> In the Data Store/VM directory, we have:
>
> ?# pwd
> ?/vmfs/volumes/T2_6045_dev_VCL_ds01/vmwarelinux-linuxtemplate2v1226-v0
> ?# ls -lh
> ?-rw--- ? ?1 root ?root ? 30.0G Jun 19 20:12
> vmwarelinux-linuxtemplate2v1226-v0-flat.vmdk ?-rw--- ? ?1 root
> ?root ? ? 527 J

FIXED, Re: VCL image capture dies, "failed to prepare vmx file"

[Michael Jinks]

For the record: the trouble was that we didn't have defined MAC
addresses in our database records for these VM's.

Adding the MAC addresses to the computer database by hand fixed the
issue.


On Tue, Jun 19, 2012 at 04:53:40PM -0500, Michael Jinks wrote:
> Hi, all.
> 
> Trying to resurrect a thread I started on Friday, subject 'Image capture
> succeeds, then "fails"'.  Mike Haudenschild helped me narrow down the
> problem but so far I haven't been able to find a cause or a fix.
> 
> I'm trying to capture my first Linux VM.  Things go along okay through
> the capture stage, but (thanks Mike) we fail when VCL tries to reserve
> and deploy the new image.
> 
> The critical message from the log appears to be:
> 
>  |17491|19:19|reload|  WARNING 
>  |17491|19:19|reload| 2012-06-19
>  14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx file 
> for vcl-linux-template-2-bak on VM host: vcl0001
>  |17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
>  |17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
>  |17491|19:19|reload| (-2) new.pm, process (line: 248)
>  |17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
>  |17491|19:19|reload| (-4) vcld, main (line: 346)
> 
> 
> Further down, I'll paste in a complete copy of the log mail I got from
> this attempt.
> 
> Our VM storage settings for this VM host are:
> 
>  Repository Path: /vmfs/volumes/nfs-vcl-templates
>  Data Store Path: /vmfs/volumes/T2_6045_dev_VCL_ds01
>  VM Path: /vmfs/volumes/T2_6045_dev_VCL_ds01
> 
> (Note Data Store and VM are explicitly the same; we've also had VM Path
> set to "(empty)" with no apparent difference in behavior.)
> 
> Both of those paths are writable on the hypervisor.  During capture, a
> VM directory appears in both locations.  In the directory under the
> Repository path, we have many vmdk files, sizes ranging from 842M to
> 1.5k, with names like:
> 
>  vmwarelinux-linuxtemplate2v1226-v0-s001.vmdk
>  vmwarelinux-linuxtemplate2v1226-v0-s002.vmdk
>  vmwarelinux-linuxtemplate2v1226-v0-s003.vmdk
>  [...]
>  vmwarelinux-linuxtemplate2v1226-v0-s016.vmdk
>  vmwarelinux-linuxtemplate2v1226-v0.vmdk
> 
> My first guess is that we have one vmdk file for each attempted capture,
> and this is a series of xdiffs on the original?
> 
> In the Data Store/VM directory, we have:
> 
>  # pwd
>  /vmfs/volumes/T2_6045_dev_VCL_ds01/vmwarelinux-linuxtemplate2v1226-v0
>  # ls -lh
>  -rw---1 root  root   30.0G Jun 19 20:12 
> vmwarelinux-linuxtemplate2v1226-v0-flat.vmdk
>  -rw---1 root  root 527 Jun 19 20:12 
> vmwarelinux-linuxtemplate2v1226-v0.vmdk
> 
> So, it seems that those locations are accessible and writable, and being
> written.  What else might cause the vmx file prep to fail?  I went to
> VMware.pm line 395, but that's a call to a subroutine, prepare_vmx(),
> which is a fairly lengthy chunk of code and doesn't appear to be
> dropping any breadcrumbs as it runs, only the failure message when it
> exits.  So stepping through that without guidance is going to be a
> challenge.
> 
> Meanwhile we've also started trying to capture a Windows image.  That's
> dying too, but with a different set of error messages.  I'll post that
> under separate cover.
> 
> Thanks for any insight.  Complete log message follows.
> -m
> 
> 
> reservation failed on vcl-linux-template-2-bak: process failed after trying 
> to load or make available
> 
> time: 2012-06-19 14:10:22
> caller: State.pm:reservation_failed(213)
> ( 0) State.pm, reservation_failed (line: 213)
> (-1) new.pm, process (line: 298)
> (-2) vcld, make_new_child (line: 568)
> (-3) vcld, main (line: 346)
> 
> management node: pilot-lab.uchicago.edu
> reservation PID: 17491
> parent vcld PID: 2491
> 
> request ID: 19
> reservation ID: 19
> request state/laststate: reload/image
> request start time: 2012-06-19 14:10:12
> request end time: 2012-06-19 14:40:12
> for imaging: no
> log ID: none
> 
> computer: vcl-linux-template-2-bak
> computer id: 7
> computer type: virtualmachine
> computer eth0 MAC address: 
> computer eth1 MAC address: 
> computer private IP address: 10.50.84.15
> computer public IP address: 10.50.84.15
> computer in block allocation: no
> provisioning module: VCL::Module::Provisioning::VMware::VMware
> 
> vm host: vcl0001
> vm host ID: 1
> vm host computer ID: 1
> vm profile: VMware ESXi - IT Services
> vm profile VM path: /vmfs/volumes/T2_6045_dev_VCL_ds01
> vm profile repository path: /vmfs/volumes/nfs-vcl-templates
> vm profile datastore path: /vmfs/volumes/T2_6045_dev_VCL_ds01
> vm profile disk type: localdisk
> 
> image: vmwarelinux-linuxtemplate2v1226-v0
> image display name: linux-template-2v12
> image ID: 26
> image revision ID: 22
> image size: 30720 MB
> use Sysprep: yes
> root access: yes
> image owner ID: 1
> image owner affiliation: Local
> image revision date creat

Re: VM MAC addressing, was Re: VCL image capture dies, "failed to prepare vmx file"

[William Robinson]

  
  
thanks.  i'm not sure i want to
blow what i have configured away.  i have a colleague working on
this with me who is pretty knowledgeable with mysql.  i'm sure i
can get him to help me fix these up.  this has also convinced me
to install phpmyadmin to simplify things in situations like
this.
  
will
  
  On 06/25/2012 10:31 AM, Mike Haudenschild wrote:

Hi Will --
  
  
  The MAC address range Mike identified below is the best to
use.  ESXi expects any manually-assigned MAC addresses to fall
within the range  00:50:56:00:yy:zz and 00:50:56:3F:yy:zz [1].
 (Note these are manually-assigned from ESXi's perspective, not
VCL's perspective.  VCL generates a VMX file with the MAC
addresses, and ESXi just accepts those addresses.)
  
  
  In 2.2.1 you must use the "multiple computers" option for
adding VMs, as that screen provides a space to enter the
starting MAC address.
  
  
  Note that VCL iterates the private and public MAC addresses
starting with the address you specify.  If you enter
00:50:56:00:00:01 as the starting address, VCL will create
computers as follows:
  
  
  VM1 private interface (eth0): 00:50:56:00:00:01
  VM1 public interface (eth1): 00:50:56:00:00:02
  VM2 eth0: 00:50:56:00:00:03
  VM2 eth1: 00:50:56:00:00:04
  etc.
  
  
  If you've already added VMs via the single-add-computer
screen in VCL, you'll need to either abandon them or use an SQL
client to edit the "computer" table and assign them MAC
addresses. [2]  This should no longer be an issue in 2.3. [3]
  
  
  [1] http://kb.vmware.com/selfservice/microsites/search.do?cmd=displayKC&externalId=507
  
[2] http://markmail.org/message/bp5ax4czlqxzyto3
  [3] http://markmail.org/message/v2vqktuefbaur2c4
  
  
  Regards,
  Mike
  
  
  

  --
  Mike Haudenschild
  Education Systems Manager
  Longsight Group
  (740) 599-5005 x809
  m...@longsight.com
  www.longsight.com
  
  
  
  On Mon, Jun 25, 2012 at 9:58 AM,
Michael Jinks 
wrote:

  Hi Will.
  
  I believe that VCL requires manually assigning MAC
  addresses when you
  generate the VM records -- you must use the multi-add page
  under "Add
  Computers", and have it assign a range of MAC addresses.
  
  Note that VMware uses two ranges of MAC addresses: one
  range for those
  it auto-generates, and one for those assigned by the user.
   I believe
  that VCL sets the MAC address of the VM at each deployment
  time, which
  would mean that you can't just let VMware generate the
  MACs and then
  list them in the database afterward; you must use
  something from the
  manual range.
  
  We chose two ranges under 00:50:56:xx:xx:xx, one for each
  of our private
  and public-facing net segments.
  
  I've read at least one doc that refers to this being the
  "easiest" way
  to set up your VM network, but after lots of experimenting
  I think it's
  actually the only way.
  
  Good luck,
  -m
  
  
  On Mon, Jun 25, 2012 at 09:13:29AM -0400, William Robinson
  wrote:
  > hi all,
  >
  > new to this list.  i have started setting up vcl here
  and clemson university and
  > this thread has touched on one of the questions i had
  in trying to do so.  do i
  > assign mac addresses based on what esxi generates?
   if not, how do i map those
  > correctly?  thanks.
  >
  > will
  >
  >
  > On 06/20/2012 10:16 AM, Josh Thompson wrote:
  > > -BEGIN PGP SIGNED MESSAGE-
  > > Hash: SHA1
  > >
  > > Michael,
  > >
  > > Unfortunately, yes, in 2.2.1, you can only set
  MAC addresses via the add
  > > multiple computers page.  You'll need to set
  them in the database.  This is
  > > fixed in 2.3.
  > >
  > > Josh
  > >
  > > On Wednesday, June 20, 2012 9:12:52 AM Michael
  Jinks wrote:
  > >> Thanks Aaron.
  > >>
  > >> Ah, yes.  It didn't occur to

Re: VM MAC addressing, was Re: VCL image capture dies, "failed to prepare vmx file"

[Mike Haudenschild]

Hi Will --

The MAC address range Mike identified below is the best to use.  ESXi
expects any manually-assigned MAC addresses to fall within the range
 00:50:56:00:yy:zz and 00:50:56:3F:yy:zz [1].  (Note these are
manually-assigned from ESXi's perspective, not VCL's perspective.  VCL
generates a VMX file with the MAC addresses, and ESXi just accepts those
addresses.)

In 2.2.1 you must use the "multiple computers" option for adding VMs, as
that screen provides a space to enter the starting MAC address.

Note that VCL iterates the private and public MAC addresses starting with
the address you specify.  If you enter 00:50:56:00:00:01 as the starting
address, VCL will create computers as follows:

VM1 private interface (eth0): 00:50:56:00:00:01
VM1 public interface (eth1): 00:50:56:00:00:02
VM2 eth0: 00:50:56:00:00:03
VM2 eth1: 00:50:56:00:00:04
etc.

If you've already added VMs via the single-add-computer screen in VCL,
you'll need to either abandon them or use an SQL client to edit the
"computer" table and assign them MAC addresses. [2]  This should no longer
be an issue in 2.3. [3]

[1]
http://kb.vmware.com/selfservice/microsites/search.do?cmd=displayKC&externalId=507
[2] http://markmail.org/message/bp5ax4czlqxzyto3
[3] http://markmail.org/message/v2vqktuefbaur2c4

Regards,
Mike

--
*Mike Haudenschild*
Education Systems Manager
Longsight Group
(740) 599-5005 x809
m...@longsight.com
www.longsight.com



On Mon, Jun 25, 2012 at 9:58 AM, Michael Jinks  wrote:

> Hi Will.
>
> I believe that VCL requires manually assigning MAC addresses when you
> generate the VM records -- you must use the multi-add page under "Add
> Computers", and have it assign a range of MAC addresses.
>
> Note that VMware uses two ranges of MAC addresses: one range for those
> it auto-generates, and one for those assigned by the user.  I believe
> that VCL sets the MAC address of the VM at each deployment time, which
> would mean that you can't just let VMware generate the MACs and then
> list them in the database afterward; you must use something from the
> manual range.
>
> We chose two ranges under 00:50:56:xx:xx:xx, one for each of our private
> and public-facing net segments.
>
> I've read at least one doc that refers to this being the "easiest" way
> to set up your VM network, but after lots of experimenting I think it's
> actually the only way.
>
> Good luck,
> -m
>
>
> On Mon, Jun 25, 2012 at 09:13:29AM -0400, William Robinson wrote:
> > hi all,
> >
> > new to this list.  i have started setting up vcl here and clemson
> university and
> > this thread has touched on one of the questions i had in trying to do
> so.  do i
> > assign mac addresses based on what esxi generates?  if not, how do i map
> those
> > correctly?  thanks.
> >
> > will
> >
> >
> > On 06/20/2012 10:16 AM, Josh Thompson wrote:
> > > -BEGIN PGP SIGNED MESSAGE-
> > > Hash: SHA1
> > >
> > > Michael,
> > >
> > > Unfortunately, yes, in 2.2.1, you can only set MAC addresses via the
> add
> > > multiple computers page.  You'll need to set them in the database.
>  This is
> > > fixed in 2.3.
> > >
> > > Josh
> > >
> > > On Wednesday, June 20, 2012 9:12:52 AM Michael Jinks wrote:
> > >> Thanks Aaron.
> > >>
> > >> Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
> > >> defined for a machine we're capturing *from*.  I figured, well, it's
> > >> running and has IP addresses...
> > >>
> > >> At any rate, I'm looking around the UI and I can't find any page that
> > >> will let me enter a MAC address for an existing computer record.  Is
> > >> this something I have to do by hand in the database?
> > >>
> > >> On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:
> > >>> Hello Michael,
> > >>>
> > >>> One thing that stands out in the log is that your MAC addresses are
> not
> > >>> defined. computer eth0 MAC address: 
> > >>> computer eth1 MAC address: 
> > >>>
> > >>> This will be needed to correctly create the vmx file and thus get an
> > >>> address from the private dhcp server that is should be running on
> your
> > >>> management node.
> > >>>
> > >>> Secondly also make sure you have the dhcp server setup on your
> > >>> management node for the private network.
> > >>>
> > >>> Aaron
> > >>>
> > >>> On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks 
> > > wrote:
> >  Hi, all.
> > 
> >  Trying to resurrect a thread I started on Friday, subject 'Image
> capture
> >  succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down
> the
> >  problem but so far I haven't been able to find a cause or a fix.
> > 
> >  I'm trying to capture my first Linux VM. ?Things go along okay
> through
> >  the capture stage, but (thanks Mike) we fail when VCL tries to
> reserve
> >  and deploy the new image.
> > 
> >  The critical message from the log appears to be:
> > 
> >  ?|17491|19:19|reload|  WARNING 
> >  ?|17491|19:19|reload| 2012-06-19
> >  ?14:10:22|17491|19:19|reload|VMware.pm:loa

Re: VM MAC addressing, was Re: VCL image capture dies, "failed to prepare vmx file"

[William Robinson]

thanks michael.

will


On 06/25/2012 09:58 AM, Michael Jinks wrote:

Hi Will.

I believe that VCL requires manually assigning MAC addresses when you
generate the VM records -- you must use the multi-add page under "Add
Computers", and have it assign a range of MAC addresses.

Note that VMware uses two ranges of MAC addresses: one range for those
it auto-generates, and one for those assigned by the user.  I believe
that VCL sets the MAC address of the VM at each deployment time, which
would mean that you can't just let VMware generate the MACs and then
list them in the database afterward; you must use something from the
manual range.

We chose two ranges under 00:50:56:xx:xx:xx, one for each of our private
and public-facing net segments.

I've read at least one doc that refers to this being the "easiest" way
to set up your VM network, but after lots of experimenting I think it's
actually the only way.

Good luck,
-m

VM MAC addressing, was Re: VCL image capture dies, "failed to prepare vmx file"

[Michael Jinks]

Hi Will.

I believe that VCL requires manually assigning MAC addresses when you
generate the VM records -- you must use the multi-add page under "Add
Computers", and have it assign a range of MAC addresses.

Note that VMware uses two ranges of MAC addresses: one range for those
it auto-generates, and one for those assigned by the user.  I believe
that VCL sets the MAC address of the VM at each deployment time, which
would mean that you can't just let VMware generate the MACs and then
list them in the database afterward; you must use something from the
manual range.

We chose two ranges under 00:50:56:xx:xx:xx, one for each of our private
and public-facing net segments.

I've read at least one doc that refers to this being the "easiest" way
to set up your VM network, but after lots of experimenting I think it's
actually the only way.

Good luck,
-m


On Mon, Jun 25, 2012 at 09:13:29AM -0400, William Robinson wrote:
> hi all,
> 
> new to this list.  i have started setting up vcl here and clemson university 
> and 
> this thread has touched on one of the questions i had in trying to do so.  do 
> i 
> assign mac addresses based on what esxi generates?  if not, how do i map 
> those 
> correctly?  thanks.
> 
> will
> 
> 
> On 06/20/2012 10:16 AM, Josh Thompson wrote:
> > -BEGIN PGP SIGNED MESSAGE-
> > Hash: SHA1
> >
> > Michael,
> >
> > Unfortunately, yes, in 2.2.1, you can only set MAC addresses via the add
> > multiple computers page.  You'll need to set them in the database.  This is
> > fixed in 2.3.
> >
> > Josh
> >
> > On Wednesday, June 20, 2012 9:12:52 AM Michael Jinks wrote:
> >> Thanks Aaron.
> >>
> >> Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
> >> defined for a machine we're capturing *from*.  I figured, well, it's
> >> running and has IP addresses...
> >>
> >> At any rate, I'm looking around the UI and I can't find any page that
> >> will let me enter a MAC address for an existing computer record.  Is
> >> this something I have to do by hand in the database?
> >>
> >> On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:
> >>> Hello Michael,
> >>>
> >>> One thing that stands out in the log is that your MAC addresses are not
> >>> defined. computer eth0 MAC address: 
> >>> computer eth1 MAC address: 
> >>>
> >>> This will be needed to correctly create the vmx file and thus get an
> >>> address from the private dhcp server that is should be running on your
> >>> management node.
> >>>
> >>> Secondly also make sure you have the dhcp server setup on your
> >>> management node for the private network.
> >>>
> >>> Aaron
> >>>
> >>> On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks 
> > wrote:
>  Hi, all.
> 
>  Trying to resurrect a thread I started on Friday, subject 'Image capture
>  succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down the
>  problem but so far I haven't been able to find a cause or a fix.
> 
>  I'm trying to capture my first Linux VM. ?Things go along okay through
>  the capture stage, but (thanks Mike) we fail when VCL tries to reserve
>  and deploy the new image.
> 
>  The critical message from the log appears to be:
> 
>  ?|17491|19:19|reload|  WARNING 
>  ?|17491|19:19|reload| 2012-06-19
>  ?14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx
>  file for vcl-linux-template-2-bak on VM host: vcl0001
>  ?|17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
>  ?|17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
>  ?|17491|19:19|reload| (-2) new.pm, process (line: 248)
>  ?|17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
>  ?|17491|19:19|reload| (-4) vcld, main (line: 346)
> 
> 
>  Further down, I'll paste in a complete copy of the log mail I got from
>  this attempt.
> 
>  Our VM storage settings for this VM host are:
> 
>  ?Repository Path: ? ? ? /vmfs/volumes/nfs-vcl-templates
>  ?Data Store Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
>  ?VM Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> 
>  (Note Data Store and VM are explicitly the same; we've also had VM Path
>  set to "(empty)" with no apparent difference in behavior.)
> 
>  Both of those paths are writable on the hypervisor. ?During capture, a
>  VM directory appears in both locations. ?In the directory under the
>  Repository path, we have many vmdk files, sizes ranging from 842M to
>  1.5k, with names like:
> 
>  ?vmwarelinux-linuxtemplate2v1226-v0-s001.vmdk
>  ?vmwarelinux-linuxtemplate2v1226-v0-s002.vmdk
>  ?vmwarelinux-linuxtemplate2v1226-v0-s003.vmdk
>  ?[...]
>  ?vmwarelinux-linuxtemplate2v1226-v0-s016.vmdk
>  ?vmwarelinux-linuxtemplate2v1226-v0.vmdk
> 
>  My first guess is that we have one vmdk file for each attempted capture,
>  and this is a series of xdiffs on the original?
> 
>  In the Data Store/VM directory, we h

Re: VCL image capture dies, "failed to prepare vmx file"

[William Robinson]

hi all,

new to this list.  i have started setting up vcl here and clemson university and 
this thread has touched on one of the questions i had in trying to do so.  do i 
assign mac addresses based on what esxi generates?  if not, how do i map those 
correctly?  thanks.


will


On 06/20/2012 10:16 AM, Josh Thompson wrote:

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Michael,

Unfortunately, yes, in 2.2.1, you can only set MAC addresses via the add
multiple computers page.  You'll need to set them in the database.  This is
fixed in 2.3.

Josh

On Wednesday, June 20, 2012 9:12:52 AM Michael Jinks wrote:

Thanks Aaron.

Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
defined for a machine we're capturing *from*.  I figured, well, it's
running and has IP addresses...

At any rate, I'm looking around the UI and I can't find any page that
will let me enter a MAC address for an existing computer record.  Is
this something I have to do by hand in the database?

On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:

Hello Michael,

One thing that stands out in the log is that your MAC addresses are not
defined. computer eth0 MAC address: 
computer eth1 MAC address: 

This will be needed to correctly create the vmx file and thus get an
address from the private dhcp server that is should be running on your
management node.

Secondly also make sure you have the dhcp server setup on your
management node for the private network.

Aaron

On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks 

wrote:

Hi, all.

Trying to resurrect a thread I started on Friday, subject 'Image capture
succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down the
problem but so far I haven't been able to find a cause or a fix.

I'm trying to capture my first Linux VM. ?Things go along okay through
the capture stage, but (thanks Mike) we fail when VCL tries to reserve
and deploy the new image.

The critical message from the log appears to be:

?|17491|19:19|reload|  WARNING 
?|17491|19:19|reload| 2012-06-19
?14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx
file for vcl-linux-template-2-bak on VM host: vcl0001
?|17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
?|17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
?|17491|19:19|reload| (-2) new.pm, process (line: 248)
?|17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
?|17491|19:19|reload| (-4) vcld, main (line: 346)


Further down, I'll paste in a complete copy of the log mail I got from
this attempt.

Our VM storage settings for this VM host are:

?Repository Path: ? ? ? /vmfs/volumes/nfs-vcl-templates
?Data Store Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
?VM Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01

(Note Data Store and VM are explicitly the same; we've also had VM Path
set to "(empty)" with no apparent difference in behavior.)

Both of those paths are writable on the hypervisor. ?During capture, a
VM directory appears in both locations. ?In the directory under the
Repository path, we have many vmdk files, sizes ranging from 842M to
1.5k, with names like:

?vmwarelinux-linuxtemplate2v1226-v0-s001.vmdk
?vmwarelinux-linuxtemplate2v1226-v0-s002.vmdk
?vmwarelinux-linuxtemplate2v1226-v0-s003.vmdk
?[...]
?vmwarelinux-linuxtemplate2v1226-v0-s016.vmdk
?vmwarelinux-linuxtemplate2v1226-v0.vmdk

My first guess is that we have one vmdk file for each attempted capture,
and this is a series of xdiffs on the original?

In the Data Store/VM directory, we have:

?# pwd
?/vmfs/volumes/T2_6045_dev_VCL_ds01/vmwarelinux-linuxtemplate2v1226-v0
?# ls -lh
?-rw--- ? ?1 root ?root ? 30.0G Jun 19 20:12
vmwarelinux-linuxtemplate2v1226-v0-flat.vmdk ?-rw--- ? ?1 root
?root ? ? 527 Jun 19 20:12 vmwarelinux-linuxtemplate2v1226-v0.vmdk

So, it seems that those locations are accessible and writable, and being
written. ?What else might cause the vmx file prep to fail? ?I went to
VMware.pm line 395, but that's a call to a subroutine, prepare_vmx(),
which is a fairly lengthy chunk of code and doesn't appear to be
dropping any breadcrumbs as it runs, only the failure message when it
exits. ?So stepping through that without guidance is going to be a
challenge.

Meanwhile we've also started trying to capture a Windows image. ?That's
dying too, but with a different set of error messages. ?I'll post that
under separate cover.

Thanks for any insight. ?Complete log message follows.
-m


reservation failed on vcl-linux-template-2-bak: process failed after
trying to load or make available
---
- time: 2012-06-19 14:10:22
caller: State.pm:reservation_failed(213)
( 0) State.pm, reservation_failed (line: 213)
(-1) new.pm, process (line: 298)
(-2) vcld, make_new_child (line: 568)
(-3) vcld, main (line: 346)

management node: pilot-lab.uchicago.edu
reservation PID: 17491
parent vcld PID: 2491

request I

Re: VCL image capture dies, "failed to prepare vmx file"

[Aaron Peeler]

ok. Then best bet is to update the computer records for the missing
mac addresses.

I typically use phpmyadmin or cmdline. I'm not sure what other tools
are available for editing mysql db's.

An issue with just deleting the computer record, is that there is also
a record in the resource table that is used to reference the
computer.id. This resource.id is used for computer groups and some
other things. If deleting only the computer record, then it would
cause other queries to fail.

Aaron

On Wed, Jun 20, 2012 at 11:16 AM, Mike Haudenschild  wrote:
> I've run into issues adding new machines that have MAC addresses matching
> "deleted" machines in 2.2.1.
>
>
>
>
> On Wed, Jun 20, 2012 at 11:10 AM, Aaron Peeler  wrote:
>>
>> I agree with Aaron C. and Mike.
>>
>> Another method, you could mark them as deleted in the VCL web
>> interface and add new ones. To delete them you would need to remove
>> them from the virtual hosts and edit the hostname so you can reuse it.
>>
>> Aaron P.
>>
>> On Wed, Jun 20, 2012 at 11:04 AM, Aaron Coburn 
>> wrote:
>> > Michael,
>> >
>> > It isn't necessary to delete the rows from 'vcl.computer'. In order to
>> > edit
>> > the MAC addresses, you only need to update the 'eth0macaddress' and
>> > 'eth1macaddress' fields in the existing records.
>> >
>> > I wouldn't recommend it, but if you really need to delete the records
>> > from
>> > the computer table, there should be no issue as long as the
>> > vcl.reservation
>> > table is empty.
>> >
>> > Aaron
>> >
>> >
>> > --
>> > Aaron Coburn
>> > Systems Administrator and Programmer
>> > Academic Technology Services, Amherst College
>> > acob...@amherst.edu
>> >
>> >
>> >
>> > On Jun 20, 2012, at 10:47 AM, Michael Jinks wrote:
>> >
>> > Okay.
>> >
>> > If I go to SQL-land and delete the record for that host from the
>> > 'computers' table, will that be enough to remove it from the system or
>> > are
>> > there other areas of the DB I need to clean out as well?
>> >
>> > (Sorry if that's a dumb question, but I've found other areas where just
>> > removing a record doesn't remove all traces of the record, like the list
>> > of virtual machines assigned to a hypervisor, so I figured I should
>> > check before I start deleting things.)
>> >
>> > Thanks all.
>> >
>> > -m
>> >
>> >
>> > On Wed, Jun 20, 2012 at 10:16:04AM -0400, Josh Thompson wrote:
>> >
>> > -BEGIN PGP SIGNED MESSAGE-
>> >
>> > Hash: SHA1
>> >
>> >
>> > Michael,
>> >
>> >
>> > Unfortunately, yes, in 2.2.1, you can only set MAC addresses via the add
>> >
>> > multiple computers page.  You'll need to set them in the database.  This
>> > is
>> >
>> > fixed in 2.3.
>> >
>> >
>> > Josh
>> >
>> >
>> > On Wednesday, June 20, 2012 9:12:52 AM Michael Jinks wrote:
>> >
>> > Thanks Aaron.
>> >
>> >
>> > Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
>> >
>> > defined for a machine we're capturing *from*.  I figured, well, it's
>> >
>> > running and has IP addresses...
>> >
>> >
>> > At any rate, I'm looking around the UI and I can't find any page that
>> >
>> > will let me enter a MAC address for an existing computer record.  Is
>> >
>> > this something I have to do by hand in the database?
>> >
>> >
>> > On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:
>> >
>> > Hello Michael,
>> >
>> >
>> > One thing that stands out in the log is that your MAC addresses are not
>> >
>> > defined. computer eth0 MAC address: 
>> >
>> > computer eth1 MAC address: 
>> >
>> >
>> > This will be needed to correctly create the vmx file and thus get an
>> >
>> > address from the private dhcp server that is should be running on your
>> >
>> > management node.
>> >
>> >
>> > Secondly also make sure you have the dhcp server setup on your
>> >
>> > management node for the private network.
>> >
>> >
>> > Aaron
>> >
>> >
>> > On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks 
>> >
>> > wrote:
>> >
>> > Hi, all.
>> >
>> >
>> > Trying to resurrect a thread I started on Friday, subject 'Image capture
>> >
>> > succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down the
>> >
>> > problem but so far I haven't been able to find a cause or a fix.
>> >
>> >
>> > I'm trying to capture my first Linux VM. ?Things go along okay through
>> >
>> > the capture stage, but (thanks Mike) we fail when VCL tries to reserve
>> >
>> > and deploy the new image.
>> >
>> >
>> > The critical message from the log appears to be:
>> >
>> >
>> > ?|17491|19:19|reload|  WARNING 
>> >
>> > ?|17491|19:19|reload| 2012-06-19
>> >
>> > ?14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx
>> >
>> > file for vcl-linux-template-2-bak on VM host: vcl0001
>> >
>> > ?|17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
>> >
>> > ?|17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
>> >
>> > ?|17491|19:19|reload| (-2) new.pm, process (line: 248)
>> >
>> > ?|17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
>> >
>> > ?|17491|19:19|reload| (-4) vcld, main (line: 346)
>> 

Re: VCL image capture dies, "failed to prepare vmx file"

[Mike Haudenschild]

I've run into issues adding new machines that have MAC addresses matching
"deleted" machines in 2.2.1.



On Wed, Jun 20, 2012 at 11:10 AM, Aaron Peeler  wrote:

> I agree with Aaron C. and Mike.
>
> Another method, you could mark them as deleted in the VCL web
> interface and add new ones. To delete them you would need to remove
> them from the virtual hosts and edit the hostname so you can reuse it.
>
> Aaron P.
>
> On Wed, Jun 20, 2012 at 11:04 AM, Aaron Coburn 
> wrote:
> > Michael,
> >
> > It isn't necessary to delete the rows from 'vcl.computer'. In order to
> edit
> > the MAC addresses, you only need to update the 'eth0macaddress' and
> > 'eth1macaddress' fields in the existing records.
> >
> > I wouldn't recommend it, but if you really need to delete the records
> from
> > the computer table, there should be no issue as long as the
> vcl.reservation
> > table is empty.
> >
> > Aaron
> >
> >
> > --
> > Aaron Coburn
> > Systems Administrator and Programmer
> > Academic Technology Services, Amherst College
> > acob...@amherst.edu
> >
> >
> >
> > On Jun 20, 2012, at 10:47 AM, Michael Jinks wrote:
> >
> > Okay.
> >
> > If I go to SQL-land and delete the record for that host from the
> > 'computers' table, will that be enough to remove it from the system or
> are
> > there other areas of the DB I need to clean out as well?
> >
> > (Sorry if that's a dumb question, but I've found other areas where just
> > removing a record doesn't remove all traces of the record, like the list
> > of virtual machines assigned to a hypervisor, so I figured I should
> > check before I start deleting things.)
> >
> > Thanks all.
> >
> > -m
> >
> >
> > On Wed, Jun 20, 2012 at 10:16:04AM -0400, Josh Thompson wrote:
> >
> > -BEGIN PGP SIGNED MESSAGE-
> >
> > Hash: SHA1
> >
> >
> > Michael,
> >
> >
> > Unfortunately, yes, in 2.2.1, you can only set MAC addresses via the add
> >
> > multiple computers page.  You'll need to set them in the database.  This
> is
> >
> > fixed in 2.3.
> >
> >
> > Josh
> >
> >
> > On Wednesday, June 20, 2012 9:12:52 AM Michael Jinks wrote:
> >
> > Thanks Aaron.
> >
> >
> > Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
> >
> > defined for a machine we're capturing *from*.  I figured, well, it's
> >
> > running and has IP addresses...
> >
> >
> > At any rate, I'm looking around the UI and I can't find any page that
> >
> > will let me enter a MAC address for an existing computer record.  Is
> >
> > this something I have to do by hand in the database?
> >
> >
> > On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:
> >
> > Hello Michael,
> >
> >
> > One thing that stands out in the log is that your MAC addresses are not
> >
> > defined. computer eth0 MAC address: 
> >
> > computer eth1 MAC address: 
> >
> >
> > This will be needed to correctly create the vmx file and thus get an
> >
> > address from the private dhcp server that is should be running on your
> >
> > management node.
> >
> >
> > Secondly also make sure you have the dhcp server setup on your
> >
> > management node for the private network.
> >
> >
> > Aaron
> >
> >
> > On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks 
> >
> > wrote:
> >
> > Hi, all.
> >
> >
> > Trying to resurrect a thread I started on Friday, subject 'Image capture
> >
> > succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down the
> >
> > problem but so far I haven't been able to find a cause or a fix.
> >
> >
> > I'm trying to capture my first Linux VM. ?Things go along okay through
> >
> > the capture stage, but (thanks Mike) we fail when VCL tries to reserve
> >
> > and deploy the new image.
> >
> >
> > The critical message from the log appears to be:
> >
> >
> > ?|17491|19:19|reload|  WARNING 
> >
> > ?|17491|19:19|reload| 2012-06-19
> >
> > ?14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx
> >
> > file for vcl-linux-template-2-bak on VM host: vcl0001
> >
> > ?|17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
> >
> > ?|17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
> >
> > ?|17491|19:19|reload| (-2) new.pm, process (line: 248)
> >
> > ?|17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
> >
> > ?|17491|19:19|reload| (-4) vcld, main (line: 346)
> >
> >
> >
> > Further down, I'll paste in a complete copy of the log mail I got from
> >
> > this attempt.
> >
> >
> > Our VM storage settings for this VM host are:
> >
> >
> > ?Repository Path: ? ? ? /vmfs/volumes/nfs-vcl-templates
> >
> > ?Data Store Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> >
> > ?VM Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> >
> >
> > (Note Data Store and VM are explicitly the same; we've also had VM Path
> >
> > set to "(empty)" with no apparent difference in behavior.)
> >
> >
> > Both of those paths are writable on the hypervisor. ?During capture, a
> >
> > VM directory appears in both locations. ?In the directory under the
> >
> > Repository path, we have many vmdk files, sizes ranging from 8

Re: VCL image capture dies, "failed to prepare vmx file"

[Aaron Peeler]

I agree with Aaron C. and Mike.

Another method, you could mark them as deleted in the VCL web
interface and add new ones. To delete them you would need to remove
them from the virtual hosts and edit the hostname so you can reuse it.

Aaron P.

On Wed, Jun 20, 2012 at 11:04 AM, Aaron Coburn  wrote:
> Michael,
>
> It isn't necessary to delete the rows from 'vcl.computer'. In order to edit
> the MAC addresses, you only need to update the 'eth0macaddress' and
> 'eth1macaddress' fields in the existing records.
>
> I wouldn't recommend it, but if you really need to delete the records from
> the computer table, there should be no issue as long as the vcl.reservation
> table is empty.
>
> Aaron
>
>
> --
> Aaron Coburn
> Systems Administrator and Programmer
> Academic Technology Services, Amherst College
> acob...@amherst.edu
>
>
>
> On Jun 20, 2012, at 10:47 AM, Michael Jinks wrote:
>
> Okay.
>
> If I go to SQL-land and delete the record for that host from the
> 'computers' table, will that be enough to remove it from the system or are
> there other areas of the DB I need to clean out as well?
>
> (Sorry if that's a dumb question, but I've found other areas where just
> removing a record doesn't remove all traces of the record, like the list
> of virtual machines assigned to a hypervisor, so I figured I should
> check before I start deleting things.)
>
> Thanks all.
>
> -m
>
>
> On Wed, Jun 20, 2012 at 10:16:04AM -0400, Josh Thompson wrote:
>
> -BEGIN PGP SIGNED MESSAGE-
>
> Hash: SHA1
>
>
> Michael,
>
>
> Unfortunately, yes, in 2.2.1, you can only set MAC addresses via the add
>
> multiple computers page.  You'll need to set them in the database.  This is
>
> fixed in 2.3.
>
>
> Josh
>
>
> On Wednesday, June 20, 2012 9:12:52 AM Michael Jinks wrote:
>
> Thanks Aaron.
>
>
> Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
>
> defined for a machine we're capturing *from*.  I figured, well, it's
>
> running and has IP addresses...
>
>
> At any rate, I'm looking around the UI and I can't find any page that
>
> will let me enter a MAC address for an existing computer record.  Is
>
> this something I have to do by hand in the database?
>
>
> On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:
>
> Hello Michael,
>
>
> One thing that stands out in the log is that your MAC addresses are not
>
> defined. computer eth0 MAC address: 
>
> computer eth1 MAC address: 
>
>
> This will be needed to correctly create the vmx file and thus get an
>
> address from the private dhcp server that is should be running on your
>
> management node.
>
>
> Secondly also make sure you have the dhcp server setup on your
>
> management node for the private network.
>
>
> Aaron
>
>
> On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks 
>
> wrote:
>
> Hi, all.
>
>
> Trying to resurrect a thread I started on Friday, subject 'Image capture
>
> succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down the
>
> problem but so far I haven't been able to find a cause or a fix.
>
>
> I'm trying to capture my first Linux VM. ?Things go along okay through
>
> the capture stage, but (thanks Mike) we fail when VCL tries to reserve
>
> and deploy the new image.
>
>
> The critical message from the log appears to be:
>
>
> ?|17491|19:19|reload|  WARNING 
>
> ?|17491|19:19|reload| 2012-06-19
>
> ?14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx
>
> file for vcl-linux-template-2-bak on VM host: vcl0001
>
> ?|17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
>
> ?|17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
>
> ?|17491|19:19|reload| (-2) new.pm, process (line: 248)
>
> ?|17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
>
> ?|17491|19:19|reload| (-4) vcld, main (line: 346)
>
>
>
> Further down, I'll paste in a complete copy of the log mail I got from
>
> this attempt.
>
>
> Our VM storage settings for this VM host are:
>
>
> ?Repository Path: ? ? ? /vmfs/volumes/nfs-vcl-templates
>
> ?Data Store Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
>
> ?VM Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
>
>
> (Note Data Store and VM are explicitly the same; we've also had VM Path
>
> set to "(empty)" with no apparent difference in behavior.)
>
>
> Both of those paths are writable on the hypervisor. ?During capture, a
>
> VM directory appears in both locations. ?In the directory under the
>
> Repository path, we have many vmdk files, sizes ranging from 842M to
>
> 1.5k, with names like:
>
>
> ?vmwarelinux-linuxtemplate2v1226-v0-s001.vmdk
>
> ?vmwarelinux-linuxtemplate2v1226-v0-s002.vmdk
>
> ?vmwarelinux-linuxtemplate2v1226-v0-s003.vmdk
>
> ?[...]
>
> ?vmwarelinux-linuxtemplate2v1226-v0-s016.vmdk
>
> ?vmwarelinux-linuxtemplate2v1226-v0.vmdk
>
>
> My first guess is that we have one vmdk file for each attempted capture,
>
> and this is a series of xdiffs on the original?
>
>
> In the Data Store/VM directory, we have:
>
>
> ?# pwd
>
> ?/vmfs/volumes/T2_6045_dev_VCL_ds01/vmwarelinux-li

Re: VCL image capture dies, "failed to prepare vmx file"

[Mike Haudenschild]

Hi Mike,

You might instead try to update the records to add the MAC addresses (both
private and public) for each machine.

http://www.tutorialspoint.com/mysql/mysql-update-query.htm

You might try removing all but a couple of the computers from the VM host,
then just updating the database records for those few to see if that
resolves the problem.

Also, on these paths:

 Repository Path:   /vmfs/volumes/nfs-vcl-templates
>  Data Store Path:   /vmfs/volumes/T2_6045_dev_VCL_ds01
>  VM Path:   /vmfs/volumes/T2_6045_dev_VCL_ds01


What is "VM Disk" set to?  Can you remind me of your storage topology?  The
repository is local to the management node, which is a Linux box?
 Obviously exported via NFS.  On the other two, I can't be 100% sure, but
maybe one of the VCL scripts is tripping up on the underscores and/or the
length of the last part of the path.  Not sure that either would matter.

Make sure that the paths that are local to the hypervisor are writable from
the management node via 'ssh -i /etc/vcl/vcl.key [vmhost]' ... I'm not
sure, but I think the VMX file is generated on the management node and then
pushed to the VM path via SCP (devs can correct me here).

I think you've covered a couple of these items in previous emails; I've
been working remotely for the past couple weeks and don't have all the
history in front of me, so apologies for anything I'm asking you to repost.

Thanks,
Mike


On Wed, Jun 20, 2012 at 10:47 AM, Michael Jinks  wrote:

> Okay.
>
> If I go to SQL-land and delete the record for that host from the
> 'computers' table, will that be enough to remove it from the system or are
> there other areas of the DB I need to clean out as well?
>
> (Sorry if that's a dumb question, but I've found other areas where just
> removing a record doesn't remove all traces of the record, like the list
> of virtual machines assigned to a hypervisor, so I figured I should
> check before I start deleting things.)
>
> Thanks all.
>
> -m
>
>
> On Wed, Jun 20, 2012 at 10:16:04AM -0400, Josh Thompson wrote:
> > -BEGIN PGP SIGNED MESSAGE-
> > Hash: SHA1
> >
> > Michael,
> >
> > Unfortunately, yes, in 2.2.1, you can only set MAC addresses via the add
> > multiple computers page.  You'll need to set them in the database.  This
> is
> > fixed in 2.3.
> >
> > Josh
> >
> > On Wednesday, June 20, 2012 9:12:52 AM Michael Jinks wrote:
> > > Thanks Aaron.
> > >
> > > Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
> > > defined for a machine we're capturing *from*.  I figured, well, it's
> > > running and has IP addresses...
> > >
> > > At any rate, I'm looking around the UI and I can't find any page that
> > > will let me enter a MAC address for an existing computer record.  Is
> > > this something I have to do by hand in the database?
> > >
> > > On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:
> > > > Hello Michael,
> > > >
> > > > One thing that stands out in the log is that your MAC addresses are
> not
> > > > defined. computer eth0 MAC address: 
> > > > computer eth1 MAC address: 
> > > >
> > > > This will be needed to correctly create the vmx file and thus get an
> > > > address from the private dhcp server that is should be running on
> your
> > > > management node.
> > > >
> > > > Secondly also make sure you have the dhcp server setup on your
> > > > management node for the private network.
> > > >
> > > > Aaron
> > > >
> > > > On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks 
> > wrote:
> > > > > Hi, all.
> > > > >
> > > > > Trying to resurrect a thread I started on Friday, subject 'Image
> capture
> > > > > succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down
> the
> > > > > problem but so far I haven't been able to find a cause or a fix.
> > > > >
> > > > > I'm trying to capture my first Linux VM. ?Things go along okay
> through
> > > > > the capture stage, but (thanks Mike) we fail when VCL tries to
> reserve
> > > > > and deploy the new image.
> > > > >
> > > > > The critical message from the log appears to be:
> > > > >
> > > > > ?|17491|19:19|reload|  WARNING 
> > > > > ?|17491|19:19|reload| 2012-06-19
> > > > > ?14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare
> vmx
> > > > > file for vcl-linux-template-2-bak on VM host: vcl0001
> > > > > ?|17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
> > > > > ?|17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
> > > > > ?|17491|19:19|reload| (-2) new.pm, process (line: 248)
> > > > > ?|17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
> > > > > ?|17491|19:19|reload| (-4) vcld, main (line: 346)
> > > > >
> > > > >
> > > > > Further down, I'll paste in a complete copy of the log mail I got
> from
> > > > > this attempt.
> > > > >
> > > > > Our VM storage settings for this VM host are:
> > > > >
> > > > > ?Repository Path: ? ? ? /vmfs/volumes/nfs-vcl-templates
> > > > > ?Data Store Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> > > > > ?VM Path: ? ? ? /vmfs

Re: VCL image capture dies, "failed to prepare vmx file"

[Michael Jinks]

Okay.

If I go to SQL-land and delete the record for that host from the
'computers' table, will that be enough to remove it from the system or are
there other areas of the DB I need to clean out as well?

(Sorry if that's a dumb question, but I've found other areas where just
removing a record doesn't remove all traces of the record, like the list
of virtual machines assigned to a hypervisor, so I figured I should
check before I start deleting things.)

Thanks all.

-m


On Wed, Jun 20, 2012 at 10:16:04AM -0400, Josh Thompson wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
> 
> Michael,
> 
> Unfortunately, yes, in 2.2.1, you can only set MAC addresses via the add 
> multiple computers page.  You'll need to set them in the database.  This is 
> fixed in 2.3.
> 
> Josh
> 
> On Wednesday, June 20, 2012 9:12:52 AM Michael Jinks wrote:
> > Thanks Aaron.
> > 
> > Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
> > defined for a machine we're capturing *from*.  I figured, well, it's
> > running and has IP addresses...
> > 
> > At any rate, I'm looking around the UI and I can't find any page that
> > will let me enter a MAC address for an existing computer record.  Is
> > this something I have to do by hand in the database?
> > 
> > On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:
> > > Hello Michael,
> > > 
> > > One thing that stands out in the log is that your MAC addresses are not
> > > defined. computer eth0 MAC address: 
> > > computer eth1 MAC address: 
> > > 
> > > This will be needed to correctly create the vmx file and thus get an
> > > address from the private dhcp server that is should be running on your
> > > management node.
> > > 
> > > Secondly also make sure you have the dhcp server setup on your
> > > management node for the private network.
> > > 
> > > Aaron
> > > 
> > > On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks  
> wrote:
> > > > Hi, all.
> > > > 
> > > > Trying to resurrect a thread I started on Friday, subject 'Image capture
> > > > succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down the
> > > > problem but so far I haven't been able to find a cause or a fix.
> > > > 
> > > > I'm trying to capture my first Linux VM. ?Things go along okay through
> > > > the capture stage, but (thanks Mike) we fail when VCL tries to reserve
> > > > and deploy the new image.
> > > > 
> > > > The critical message from the log appears to be:
> > > > 
> > > > ?|17491|19:19|reload|  WARNING 
> > > > ?|17491|19:19|reload| 2012-06-19
> > > > ?14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx
> > > > file for vcl-linux-template-2-bak on VM host: vcl0001
> > > > ?|17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
> > > > ?|17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
> > > > ?|17491|19:19|reload| (-2) new.pm, process (line: 248)
> > > > ?|17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
> > > > ?|17491|19:19|reload| (-4) vcld, main (line: 346)
> > > > 
> > > > 
> > > > Further down, I'll paste in a complete copy of the log mail I got from
> > > > this attempt.
> > > > 
> > > > Our VM storage settings for this VM host are:
> > > > 
> > > > ?Repository Path: ? ? ? /vmfs/volumes/nfs-vcl-templates
> > > > ?Data Store Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> > > > ?VM Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> > > > 
> > > > (Note Data Store and VM are explicitly the same; we've also had VM Path
> > > > set to "(empty)" with no apparent difference in behavior.)
> > > > 
> > > > Both of those paths are writable on the hypervisor. ?During capture, a
> > > > VM directory appears in both locations. ?In the directory under the
> > > > Repository path, we have many vmdk files, sizes ranging from 842M to
> > > > 1.5k, with names like:
> > > > 
> > > > ?vmwarelinux-linuxtemplate2v1226-v0-s001.vmdk
> > > > ?vmwarelinux-linuxtemplate2v1226-v0-s002.vmdk
> > > > ?vmwarelinux-linuxtemplate2v1226-v0-s003.vmdk
> > > > ?[...]
> > > > ?vmwarelinux-linuxtemplate2v1226-v0-s016.vmdk
> > > > ?vmwarelinux-linuxtemplate2v1226-v0.vmdk
> > > > 
> > > > My first guess is that we have one vmdk file for each attempted capture,
> > > > and this is a series of xdiffs on the original?
> > > > 
> > > > In the Data Store/VM directory, we have:
> > > > 
> > > > ?# pwd
> > > > ?/vmfs/volumes/T2_6045_dev_VCL_ds01/vmwarelinux-linuxtemplate2v1226-v0
> > > > ?# ls -lh
> > > > ?-rw--- ? ?1 root ?root ? 30.0G Jun 19 20:12
> > > > vmwarelinux-linuxtemplate2v1226-v0-flat.vmdk ?-rw--- ? ?1 root
> > > > ?root ? ? 527 Jun 19 20:12 vmwarelinux-linuxtemplate2v1226-v0.vmdk
> > > > 
> > > > So, it seems that those locations are accessible and writable, and being
> > > > written. ?What else might cause the vmx file prep to fail? ?I went to
> > > > VMware.pm line 395, but that's a call to a subroutine, prepare_vmx(),
> > > > which is a fairly lengthy chunk of code and doesn't appear to be
> > > > dropping any breadcrumbs as it runs, only t

Re: VCL image capture dies, "failed to prepare vmx file"

[Josh Thompson]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Michael,

Unfortunately, yes, in 2.2.1, you can only set MAC addresses via the add 
multiple computers page.  You'll need to set them in the database.  This is 
fixed in 2.3.

Josh

On Wednesday, June 20, 2012 9:12:52 AM Michael Jinks wrote:
> Thanks Aaron.
> 
> Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
> defined for a machine we're capturing *from*.  I figured, well, it's
> running and has IP addresses...
> 
> At any rate, I'm looking around the UI and I can't find any page that
> will let me enter a MAC address for an existing computer record.  Is
> this something I have to do by hand in the database?
> 
> On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:
> > Hello Michael,
> > 
> > One thing that stands out in the log is that your MAC addresses are not
> > defined. computer eth0 MAC address: 
> > computer eth1 MAC address: 
> > 
> > This will be needed to correctly create the vmx file and thus get an
> > address from the private dhcp server that is should be running on your
> > management node.
> > 
> > Secondly also make sure you have the dhcp server setup on your
> > management node for the private network.
> > 
> > Aaron
> > 
> > On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks  
wrote:
> > > Hi, all.
> > > 
> > > Trying to resurrect a thread I started on Friday, subject 'Image capture
> > > succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down the
> > > problem but so far I haven't been able to find a cause or a fix.
> > > 
> > > I'm trying to capture my first Linux VM. ?Things go along okay through
> > > the capture stage, but (thanks Mike) we fail when VCL tries to reserve
> > > and deploy the new image.
> > > 
> > > The critical message from the log appears to be:
> > > 
> > > ?|17491|19:19|reload|  WARNING 
> > > ?|17491|19:19|reload| 2012-06-19
> > > ?14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx
> > > file for vcl-linux-template-2-bak on VM host: vcl0001
> > > ?|17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
> > > ?|17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
> > > ?|17491|19:19|reload| (-2) new.pm, process (line: 248)
> > > ?|17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
> > > ?|17491|19:19|reload| (-4) vcld, main (line: 346)
> > > 
> > > 
> > > Further down, I'll paste in a complete copy of the log mail I got from
> > > this attempt.
> > > 
> > > Our VM storage settings for this VM host are:
> > > 
> > > ?Repository Path: ? ? ? /vmfs/volumes/nfs-vcl-templates
> > > ?Data Store Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> > > ?VM Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> > > 
> > > (Note Data Store and VM are explicitly the same; we've also had VM Path
> > > set to "(empty)" with no apparent difference in behavior.)
> > > 
> > > Both of those paths are writable on the hypervisor. ?During capture, a
> > > VM directory appears in both locations. ?In the directory under the
> > > Repository path, we have many vmdk files, sizes ranging from 842M to
> > > 1.5k, with names like:
> > > 
> > > ?vmwarelinux-linuxtemplate2v1226-v0-s001.vmdk
> > > ?vmwarelinux-linuxtemplate2v1226-v0-s002.vmdk
> > > ?vmwarelinux-linuxtemplate2v1226-v0-s003.vmdk
> > > ?[...]
> > > ?vmwarelinux-linuxtemplate2v1226-v0-s016.vmdk
> > > ?vmwarelinux-linuxtemplate2v1226-v0.vmdk
> > > 
> > > My first guess is that we have one vmdk file for each attempted capture,
> > > and this is a series of xdiffs on the original?
> > > 
> > > In the Data Store/VM directory, we have:
> > > 
> > > ?# pwd
> > > ?/vmfs/volumes/T2_6045_dev_VCL_ds01/vmwarelinux-linuxtemplate2v1226-v0
> > > ?# ls -lh
> > > ?-rw--- ? ?1 root ?root ? 30.0G Jun 19 20:12
> > > vmwarelinux-linuxtemplate2v1226-v0-flat.vmdk ?-rw--- ? ?1 root
> > > ?root ? ? 527 Jun 19 20:12 vmwarelinux-linuxtemplate2v1226-v0.vmdk
> > > 
> > > So, it seems that those locations are accessible and writable, and being
> > > written. ?What else might cause the vmx file prep to fail? ?I went to
> > > VMware.pm line 395, but that's a call to a subroutine, prepare_vmx(),
> > > which is a fairly lengthy chunk of code and doesn't appear to be
> > > dropping any breadcrumbs as it runs, only the failure message when it
> > > exits. ?So stepping through that without guidance is going to be a
> > > challenge.
> > > 
> > > Meanwhile we've also started trying to capture a Windows image. ?That's
> > > dying too, but with a different set of error messages. ?I'll post that
> > > under separate cover.
> > > 
> > > Thanks for any insight. ?Complete log message follows.
> > > -m
> > > 
> > > 
> > > reservation failed on vcl-linux-template-2-bak: process failed after
> > > trying to load or make available
> > > ---
> > > - time: 2012-06-19 14:10:22
> > > caller: State.pm:reservation_failed(213)
> > > ( 0) State.pm, reservation_failed (line: 213)
> > > (-1) new.pm, proce

Re: VCL image capture dies, "failed to prepare vmx file"

[Michael Jinks]

Thanks Aaron.

Ah, yes.  It didn't occur to me that we'd have to have MAC addresses
defined for a machine we're capturing *from*.  I figured, well, it's
running and has IP addresses...

At any rate, I'm looking around the UI and I can't find any page that
will let me enter a MAC address for an existing computer record.  Is
this something I have to do by hand in the database?



On Wed, Jun 20, 2012 at 08:46:40AM -0400, Aaron Peeler wrote:
> Hello Michael,
> 
> One thing that stands out in the log is that your MAC addresses are not 
> defined.
> computer eth0 MAC address: 
> computer eth1 MAC address: 
> 
> This will be needed to correctly create the vmx file and thus get an
> address from the private dhcp server that is should be running on your
> management node.
> 
> Secondly also make sure you have the dhcp server setup on your
> management node for the private network.
> 
> Aaron
> 
> 
> On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks  wrote:
> > Hi, all.
> >
> > Trying to resurrect a thread I started on Friday, subject 'Image capture
> > succeeds, then "fails"'. ?Mike Haudenschild helped me narrow down the
> > problem but so far I haven't been able to find a cause or a fix.
> >
> > I'm trying to capture my first Linux VM. ?Things go along okay through
> > the capture stage, but (thanks Mike) we fail when VCL tries to reserve
> > and deploy the new image.
> >
> > The critical message from the log appears to be:
> >
> > ?|17491|19:19|reload|  WARNING 
> > ?|17491|19:19|reload| 2012-06-19
> > ?14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx file 
> > for vcl-linux-template-2-bak on VM host: vcl0001
> > ?|17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
> > ?|17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
> > ?|17491|19:19|reload| (-2) new.pm, process (line: 248)
> > ?|17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
> > ?|17491|19:19|reload| (-4) vcld, main (line: 346)
> >
> >
> > Further down, I'll paste in a complete copy of the log mail I got from
> > this attempt.
> >
> > Our VM storage settings for this VM host are:
> >
> > ?Repository Path: ? ? ? /vmfs/volumes/nfs-vcl-templates
> > ?Data Store Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> > ?VM Path: ? ? ? /vmfs/volumes/T2_6045_dev_VCL_ds01
> >
> > (Note Data Store and VM are explicitly the same; we've also had VM Path
> > set to "(empty)" with no apparent difference in behavior.)
> >
> > Both of those paths are writable on the hypervisor. ?During capture, a
> > VM directory appears in both locations. ?In the directory under the
> > Repository path, we have many vmdk files, sizes ranging from 842M to
> > 1.5k, with names like:
> >
> > ?vmwarelinux-linuxtemplate2v1226-v0-s001.vmdk
> > ?vmwarelinux-linuxtemplate2v1226-v0-s002.vmdk
> > ?vmwarelinux-linuxtemplate2v1226-v0-s003.vmdk
> > ?[...]
> > ?vmwarelinux-linuxtemplate2v1226-v0-s016.vmdk
> > ?vmwarelinux-linuxtemplate2v1226-v0.vmdk
> >
> > My first guess is that we have one vmdk file for each attempted capture,
> > and this is a series of xdiffs on the original?
> >
> > In the Data Store/VM directory, we have:
> >
> > ?# pwd
> > ?/vmfs/volumes/T2_6045_dev_VCL_ds01/vmwarelinux-linuxtemplate2v1226-v0
> > ?# ls -lh
> > ?-rw--- ? ?1 root ?root ? 30.0G Jun 19 20:12 
> > vmwarelinux-linuxtemplate2v1226-v0-flat.vmdk
> > ?-rw--- ? ?1 root ?root ? ? 527 Jun 19 20:12 
> > vmwarelinux-linuxtemplate2v1226-v0.vmdk
> >
> > So, it seems that those locations are accessible and writable, and being
> > written. ?What else might cause the vmx file prep to fail? ?I went to
> > VMware.pm line 395, but that's a call to a subroutine, prepare_vmx(),
> > which is a fairly lengthy chunk of code and doesn't appear to be
> > dropping any breadcrumbs as it runs, only the failure message when it
> > exits. ?So stepping through that without guidance is going to be a
> > challenge.
> >
> > Meanwhile we've also started trying to capture a Windows image. ?That's
> > dying too, but with a different set of error messages. ?I'll post that
> > under separate cover.
> >
> > Thanks for any insight. ?Complete log message follows.
> > -m
> >
> >
> > reservation failed on vcl-linux-template-2-bak: process failed after trying 
> > to load or make available
> > 
> > time: 2012-06-19 14:10:22
> > caller: State.pm:reservation_failed(213)
> > ( 0) State.pm, reservation_failed (line: 213)
> > (-1) new.pm, process (line: 298)
> > (-2) vcld, make_new_child (line: 568)
> > (-3) vcld, main (line: 346)
> > 
> > management node: pilot-lab.uchicago.edu
> > reservation PID: 17491
> > parent vcld PID: 2491
> >
> > request ID: 19
> > reservation ID: 19
> > request state/laststate: reload/image
> > request start time: 2012-06-19 14:10:12
> > request end time: 2012-06-19 14:40:12
> > for imaging: no
> > log ID: none
> >
> > computer: vcl-linux-template-2

Re: VCL image capture dies, "failed to prepare vmx file"

[Aaron Peeler]

Hello Michael,

One thing that stands out in the log is that your MAC addresses are not defined.
computer eth0 MAC address: 
computer eth1 MAC address: 

This will be needed to correctly create the vmx file and thus get an
address from the private dhcp server that is should be running on your
management node.

Secondly also make sure you have the dhcp server setup on your
management node for the private network.

Aaron


On Tue, Jun 19, 2012 at 5:53 PM, Michael Jinks  wrote:
> Hi, all.
>
> Trying to resurrect a thread I started on Friday, subject 'Image capture
> succeeds, then "fails"'.  Mike Haudenschild helped me narrow down the
> problem but so far I haven't been able to find a cause or a fix.
>
> I'm trying to capture my first Linux VM.  Things go along okay through
> the capture stage, but (thanks Mike) we fail when VCL tries to reserve
> and deploy the new image.
>
> The critical message from the log appears to be:
>
>  |17491|19:19|reload|  WARNING 
>  |17491|19:19|reload| 2012-06-19
>  14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx file 
> for vcl-linux-template-2-bak on VM host: vcl0001
>  |17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
>  |17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
>  |17491|19:19|reload| (-2) new.pm, process (line: 248)
>  |17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
>  |17491|19:19|reload| (-4) vcld, main (line: 346)
>
>
> Further down, I'll paste in a complete copy of the log mail I got from
> this attempt.
>
> Our VM storage settings for this VM host are:
>
>  Repository Path:       /vmfs/volumes/nfs-vcl-templates
>  Data Store Path:       /vmfs/volumes/T2_6045_dev_VCL_ds01
>  VM Path:       /vmfs/volumes/T2_6045_dev_VCL_ds01
>
> (Note Data Store and VM are explicitly the same; we've also had VM Path
> set to "(empty)" with no apparent difference in behavior.)
>
> Both of those paths are writable on the hypervisor.  During capture, a
> VM directory appears in both locations.  In the directory under the
> Repository path, we have many vmdk files, sizes ranging from 842M to
> 1.5k, with names like:
>
>  vmwarelinux-linuxtemplate2v1226-v0-s001.vmdk
>  vmwarelinux-linuxtemplate2v1226-v0-s002.vmdk
>  vmwarelinux-linuxtemplate2v1226-v0-s003.vmdk
>  [...]
>  vmwarelinux-linuxtemplate2v1226-v0-s016.vmdk
>  vmwarelinux-linuxtemplate2v1226-v0.vmdk
>
> My first guess is that we have one vmdk file for each attempted capture,
> and this is a series of xdiffs on the original?
>
> In the Data Store/VM directory, we have:
>
>  # pwd
>  /vmfs/volumes/T2_6045_dev_VCL_ds01/vmwarelinux-linuxtemplate2v1226-v0
>  # ls -lh
>  -rw---    1 root  root   30.0G Jun 19 20:12 
> vmwarelinux-linuxtemplate2v1226-v0-flat.vmdk
>  -rw---    1 root  root     527 Jun 19 20:12 
> vmwarelinux-linuxtemplate2v1226-v0.vmdk
>
> So, it seems that those locations are accessible and writable, and being
> written.  What else might cause the vmx file prep to fail?  I went to
> VMware.pm line 395, but that's a call to a subroutine, prepare_vmx(),
> which is a fairly lengthy chunk of code and doesn't appear to be
> dropping any breadcrumbs as it runs, only the failure message when it
> exits.  So stepping through that without guidance is going to be a
> challenge.
>
> Meanwhile we've also started trying to capture a Windows image.  That's
> dying too, but with a different set of error messages.  I'll post that
> under separate cover.
>
> Thanks for any insight.  Complete log message follows.
> -m
>
>
> reservation failed on vcl-linux-template-2-bak: process failed after trying 
> to load or make available
> 
> time: 2012-06-19 14:10:22
> caller: State.pm:reservation_failed(213)
> ( 0) State.pm, reservation_failed (line: 213)
> (-1) new.pm, process (line: 298)
> (-2) vcld, make_new_child (line: 568)
> (-3) vcld, main (line: 346)
> 
> management node: pilot-lab.uchicago.edu
> reservation PID: 17491
> parent vcld PID: 2491
>
> request ID: 19
> reservation ID: 19
> request state/laststate: reload/image
> request start time: 2012-06-19 14:10:12
> request end time: 2012-06-19 14:40:12
> for imaging: no
> log ID: none
>
> computer: vcl-linux-template-2-bak
> computer id: 7
> computer type: virtualmachine
> computer eth0 MAC address: 
> computer eth1 MAC address: 
> computer private IP address: 10.50.84.15
> computer public IP address: 10.50.84.15
> computer in block allocation: no
> provisioning module: VCL::Module::Provisioning::VMware::VMware
>
> vm host: vcl0001
> vm host ID: 1
> vm host computer ID: 1
> vm profile: VMware ESXi - IT Services
> vm profile VM path: /vmfs/volumes/T2_6045_dev_VCL_ds01
> vm profile repository path: /vmfs/volumes/nfs-vcl-templates
> vm profile datastore path: /vmfs/volumes/T2_6045_dev_VCL_ds01
> vm profile disk type: localdisk
>
> image: vmwarelinux-linuxtemplate2v1226-v0
> image displa

VCL image capture dies, "failed to prepare vmx file"

[Michael Jinks]

Hi, all.

Trying to resurrect a thread I started on Friday, subject 'Image capture
succeeds, then "fails"'.  Mike Haudenschild helped me narrow down the
problem but so far I haven't been able to find a cause or a fix.

I'm trying to capture my first Linux VM.  Things go along okay through
the capture stage, but (thanks Mike) we fail when VCL tries to reserve
and deploy the new image.

The critical message from the log appears to be:

 |17491|19:19|reload|  WARNING 
 |17491|19:19|reload| 2012-06-19
 14:10:22|17491|19:19|reload|VMware.pm:load(395)|failed to prepare vmx file for 
vcl-linux-template-2-bak on VM host: vcl0001
 |17491|19:19|reload| ( 0) VMware.pm, load (line: 395)
 |17491|19:19|reload| (-1) new.pm, reload_image (line: 618)
 |17491|19:19|reload| (-2) new.pm, process (line: 248)
 |17491|19:19|reload| (-3) vcld, make_new_child (line: 568)
 |17491|19:19|reload| (-4) vcld, main (line: 346)


Further down, I'll paste in a complete copy of the log mail I got from
this attempt.

Our VM storage settings for this VM host are:

 Repository Path:   /vmfs/volumes/nfs-vcl-templates
 Data Store Path:   /vmfs/volumes/T2_6045_dev_VCL_ds01
 VM Path:   /vmfs/volumes/T2_6045_dev_VCL_ds01

(Note Data Store and VM are explicitly the same; we've also had VM Path
set to "(empty)" with no apparent difference in behavior.)

Both of those paths are writable on the hypervisor.  During capture, a
VM directory appears in both locations.  In the directory under the
Repository path, we have many vmdk files, sizes ranging from 842M to
1.5k, with names like:

 vmwarelinux-linuxtemplate2v1226-v0-s001.vmdk
 vmwarelinux-linuxtemplate2v1226-v0-s002.vmdk
 vmwarelinux-linuxtemplate2v1226-v0-s003.vmdk
 [...]
 vmwarelinux-linuxtemplate2v1226-v0-s016.vmdk
 vmwarelinux-linuxtemplate2v1226-v0.vmdk

My first guess is that we have one vmdk file for each attempted capture,
and this is a series of xdiffs on the original?

In the Data Store/VM directory, we have:

 # pwd
 /vmfs/volumes/T2_6045_dev_VCL_ds01/vmwarelinux-linuxtemplate2v1226-v0
 # ls -lh
 -rw---1 root  root   30.0G Jun 19 20:12 
vmwarelinux-linuxtemplate2v1226-v0-flat.vmdk
 -rw---1 root  root 527 Jun 19 20:12 
vmwarelinux-linuxtemplate2v1226-v0.vmdk

So, it seems that those locations are accessible and writable, and being
written.  What else might cause the vmx file prep to fail?  I went to
VMware.pm line 395, but that's a call to a subroutine, prepare_vmx(),
which is a fairly lengthy chunk of code and doesn't appear to be
dropping any breadcrumbs as it runs, only the failure message when it
exits.  So stepping through that without guidance is going to be a
challenge.

Meanwhile we've also started trying to capture a Windows image.  That's
dying too, but with a different set of error messages.  I'll post that
under separate cover.

Thanks for any insight.  Complete log message follows.
-m


reservation failed on vcl-linux-template-2-bak: process failed after trying to 
load or make available

time: 2012-06-19 14:10:22
caller: State.pm:reservation_failed(213)
( 0) State.pm, reservation_failed (line: 213)
(-1) new.pm, process (line: 298)
(-2) vcld, make_new_child (line: 568)
(-3) vcld, main (line: 346)

management node: pilot-lab.uchicago.edu
reservation PID: 17491
parent vcld PID: 2491

request ID: 19
reservation ID: 19
request state/laststate: reload/image
request start time: 2012-06-19 14:10:12
request end time: 2012-06-19 14:40:12
for imaging: no
log ID: none

computer: vcl-linux-template-2-bak
computer id: 7
computer type: virtualmachine
computer eth0 MAC address: 
computer eth1 MAC address: 
computer private IP address: 10.50.84.15
computer public IP address: 10.50.84.15
computer in block allocation: no
provisioning module: VCL::Module::Provisioning::VMware::VMware

vm host: vcl0001
vm host ID: 1
vm host computer ID: 1
vm profile: VMware ESXi - IT Services
vm profile VM path: /vmfs/volumes/T2_6045_dev_VCL_ds01
vm profile repository path: /vmfs/volumes/nfs-vcl-templates
vm profile datastore path: /vmfs/volumes/T2_6045_dev_VCL_ds01
vm profile disk type: localdisk

image: vmwarelinux-linuxtemplate2v1226-v0
image display name: linux-template-2v12
image ID: 26
image revision ID: 22
image size: 30720 MB
use Sysprep: yes
root access: yes
image owner ID: 1
image owner affiliation: Local
image revision date created: 2012-06-19 14:08:20
image revision production: yes
OS module: VCL::Module::OS::Linux

user: vclreload
user name: vcl reload
user ID: 2
user affiliation: Local

RECENT LOG ENTRIES FOR THIS PROCESS:
2012-06-19 
14:10:20|17491|19:19|reload|VMware.pm:get_image_size_bytes(3841)|retrieved the 
size of the image from the datastore on the VM host: 32,212,255,247
2012-06-19 
14:10:20|17491|19:19|reload|VMware.pm:get_image_size_bytes