Re: Windows capture fails
Hi Michael and team, Could you share your vmware host server and vmware profile information that you have stored in your setup? Based on the log output there might be some missing information or it's not setup correctly. Since this is a public list, if there are any public IP address or anything important feel free to obscure that info or send to me privately. Thanks, Aaron On Wed, Jun 20, 2012 at 10:01 AM, Michael Jinks mji...@uchicago.edu wrote: We're using 2.2.1. When I downloaded the package it was the only release I could find without going to svn. On Wed, Jun 20, 2012 at 08:48:01AM -0400, Aaron Peeler wrote: Hello Michael, Which version are you using? VCL 2.2.1 or the pre-release of 2.3? The execute subroutine is not implemented is concerning. Thanks, Aaron On Tue, Jun 19, 2012 at 7:53 PM, Michael Jinks mji...@uchicago.edu wrote: Hi List. ?As threatened, here's the tale of our attempt to capture our first Windows image. ?As with our Linux capture, it appears that the VM disk image is being copied into our Data and Repository paths, but something goes wrong later in the process, but the warnings we find in the log are different for our Windows image than for the Linux one. As before, I'll paste the full log from the PROBLEM mail below, but it looks like this is significant: ?|19609|20:20|image| WARNING ?|19609|20:20|image| 2012-06-19 14:36:58|19609|20:20|image|VIM_SSH.pm:initialize(112)|required VM host OS subroutine is not implemented: execute ?|19609|20:20|image| ( 0) VIM_SSH.pm, initialize (line: 112) ?|19609|20:20|image| (-1) Module.pm, new (line: 207) ?|19609|20:20|image| (-2) VMware.pm, (eval) (line: 1303) ?|19609|20:20|image| (-3) VMware.pm, get_vmhost_api_object (line: 1303) ?|19609|20:20|image| (-4) VMware.pm, initialize (line: 278) ?|19609|20:20|image| (-5) Module.pm, new (line: 207) ?2012-06-19 14:36:58|19609|20:20|image|VMware.pm:get_vmhost_api_object(1313)|API object could not be created: VCL::Module::Provisioning::VMware::VIM_SSH This looks to me like there must be something missing, maybe within the Windows VM itself? ?Did we leave out a package somewhere? Thanks, -m -- Michael Jinks :: mji...@uchicago.edu University of Chicago IT Services Complete log message follows: VCL::image object could not be created and initialized time: 2012-06-19 14:36:58 caller: vcld:make_new_child(571) ( 0) vcld, make_new_child (line: 571) (-1) vcld, main (line: 346) management node: pilot-lab.uchicago.edu reservation PID: 19609 parent vcld PID: 2491 request ID: 20 reservation ID: 20 request state/laststate: image/image request start time: 2012-06-19 14:36:50 request end time: 2012-06-19 15:36:50 for imaging: no log ID: none computer: vcl-win7-test computer id: 8 computer type: virtualmachine computer eth0 MAC address: undefined computer eth1 MAC address: undefined computer private IP address: 10.50.84.16 computer public IP address: 10.50.84.16 computer in block allocation: no provisioning module: VCL::Module::Provisioning::VMware::VMware vm host: undefined vm host ID: undefined vm host computer ID: undefined vm profile: undefined vm profile VM path: undefined vm profile repository path: undefined vm profile datastore path: undefined vm profile disk type: undefined image: vmwarewin7-win7vmimage27-v0 image display name: win7-vm-image image ID: 27 image revision ID: 23 image size: 1450 MB use Sysprep: no root access: yes image owner ID: 1 image owner affiliation: Local image revision date created: 2012-06-19 14:36:50 image revision production: yes OS module: VCL::Module::OS::Windows::Version_6::7 user: admin user name: vcl admin user ID: 1 user affiliation: Local RECENT LOG ENTRIES FOR THIS PROCESS: |19609|20:20|image| (-1) Autoload.pm, __ANON__ (line: 80) |19609|20:20|image| (-2) VMware.pm, get_vmhost_datastructure (line: 1161) |19609|20:20|image| (-3) VMware.pm, get_vmhost_api_object (line: 1282) |19609|20:20|image| (-4) VMware.pm, initialize (line: 278) |19609|20:20|image| (-5) Module.pm, new (line: 207) |19609|20:20|image| WARNING |19609|20:20|image| 2012-06-19 14:36:58|19609|20:20|image|DataStructure.pm:_automethod(812)|corresponding data has not been initialized for get_vmhost_profile_image_id: $self-request_data-{reservation}{20}{computer}{vmhost}{vmprofile}{imageid} |19609|20:20|image| ( 0) DataStructure.pm, _automethod (line: 812) |19609|20:20|image| (-1) Autoload.pm, __ANON__ (line: 80) |19609|20:20|image| (-2) VMware.pm, get_vmhost_datastructure (line: 1162) |19609|20:20|image| (-3) VMware.pm, get_vmhost_api_object (line: 1282)
Re: libvirt/KVM
I don't have a libvirtlinux OS entry. I use vmwarelinux for KVM-based Linux images. This is what should be set when you capture a base image using vcld --setup. There is no difference among OS modules which point to the same Perl package path. The OS table is actually somewhat of a mess right now. Eventually it will be collapsed/normalized down so the is a single entry for all Linux images, otherwise we will end up with n-squared entries (OS's and provisioning types). Issue VCL-566 addresses this. -Andy On Wed, Jun 20, 2012 at 8:21 PM, Dmitri Chebotarov dcheb...@gmu.edu wrote: Andy Success! Thank you for your help. Correct, I'm working with KVM-based Linux image. I was able to make it work with a slight changes. Do you have any entries in vcl.OS table with installtype='libvirt'? The only way I'm able to make reservations is by adding new entry into vcl.OS table and then setting image.osid to the id of the new entry (49 in my case): This is the entry I've added: select * from OS where installtype='libvirt'; ++--+--+---+-+---+--+ | id | name | prettyname | type | installtype | sourcepath | moduleid | ++--+--+---+-+---+--+ | 49 | libvirtlinux | CentOS 5/6 (libvirt) | linux | libvirt | vmware_images | 5 | ++--+--+---+-+---+--+ Thanks. -- Dmitri Chebotarov Virtual Computing Lab Systems Engineer, TSD - Ent Servers Messaging 223 Aquia Building, Ffx, MSN: 1B5 Phone: (703) 993-6175 Fax: (703) 993-3404 On Wednesday, June 20, 2012 at 15:43 , Andy Kurth wrote: I'm confused about what you're trying to load. Is centos5image-kvmlinuxv4105-v0 a bare metal image? The reservation info you sent earlier had this image being loaded on the virtual machine sandbox-vcl-19. Assuming you have a Linux VM image, its OSid should be set to vmwarelinux, which should be pointing to VCL::Module::OS::Linux. The computer.provisioningid value of the VM should be set the libvirt entry in the provisioning table, which should be pointing to VCL::Module::Provisioning::libvirt. -Andy On Wed, Jun 20, 2012 at 2:57 PM, Dmitri Chebotarov dcheb...@gmu.edu (mailto:dcheb...@gmu.edu) wrote: I think I'm missing something here. I fail to understand how KVM-based image is linked to libvirt.pm/KVM.pm modules in this case (and if it should). Once I set image.osid to 36, (36 is os.id for vmwarelinux, module 5 - VCL::Module::OS::Linux), vcld is trying to load KVM-based image to a computer with Provisioning Engine set to VMware and using a VMware host. It fails. In this case (image.OSid = 36) there is no link between KVM image and libvirt computers/KVM hosts. Thanks. --- Dmitri Chebotarov Virtual Computing Lab Systems Engineer, TSD - Ent Servers Messaging 223 Aquia Building, Ffx, MSN: 1B5 Phone: (703) 993-6175 Fax: (703) 993-3404 On Wednesday, June 20, 2012 at 13:06 , Andy Kurth wrote: image.OSid should be set to the id that corresponds to vmwarelinux. OS.moduleid should point to the os_linux module table entry, which has module.perlpackagepath = VCL::Module::OS::Linux. -Andy On Wed, Jun 20, 2012 at 12:30 PM, Dmitri Chebotarov dcheb...@gmu.edu (mailto:dcheb...@gmu.edu) wrote: Andy It's pointing to VCL::Module::Provisioning::libvirt This is a result of my earlier attempt to troubleshoot the issue. I've added new OSid, moduleid. The original image.osid was 29, which is CentOS Bare Metal, processed by VCL::Module::OS::Linux. With image.osid set to 29 VCL was trying to use xCAT module to process KVM reservations. At this moment I don't have any xCAT (bare metal) computers available and once I select the [KVM] image for reservation I get popup asking to select available time for reservation, which means that I don't have available resources to support this reservation. When I've added new OSid and moduleid (pointing to VCL::Module::Provisioning::libvirt), I'm able to start reservation, VCLd finds available computers (which are configured to use libvirt engine) and attempts to make reservation on correct KVM host, but fails with error attached in previous email. What should image.osid, moduleid be set to? -- Dmitri Chebotarov Virtual Computing Lab Systems Engineer, TSD - Ent Servers Messaging 223 Aquia Building, Ffx, MSN: 1B5 Phone: (703) 993-6175 Fax: (703) 993-3404 On Wednesday, June 20, 2012 at 11:52 , Andy Kurth wrote: For some reason the OS module for the image being loaded is showing as VCL::Module::Provisioning::libvirt. Check which module centos5image-kvmlinuxv4105-v0 is configured to use in the database: image.OSid -- OS.moduleid --
VCL: How to preconfigure deployed images?
We're still trying to work out image capture on our VCL setup, but I'm wondering about an issue I think we'll encounter fairly soon and how others are addressing it. On our Linux image, /etc/ssh/sshd_config (the private-facing sshd configuration) has a ListenAddress hard-coded to the private IP that the image has before capture. Obviously, when we deploy that image to a new instance, that IP address will be wrong, and sshd will fail to bind. The same issue exists for the public-facing instance. We can't set either one to the 0.0.0.0 wildcard, because then the first sshd to come up will prevent the second from binding. All of that has been done according to the VCL docs, so I'm sure this is an issue that everybody faces, but what is the cure? We do have a confugration manager (puppet) which we can use to perform per-host tweaks, but the host has to be on the wire first, and I'm worried that a failure to have sshd running at boot time will cause the deployment of the image to fail. Maybe that's a non-issue. Anyhow... How are other sites dealing with this? Is there something in VCL that takes care of this that I've just missed? Thanks, -m -- Michael Jinks :: mji...@uchicago.edu University of Chicago IT Services
VCL: deployment to VMware, NICs disconnected
Our story so far: I have Linux base image, which I've been able to capture, and the capture process deploys the image to a new VM and boots it. So far so good. But the network isn't working. When I go to the VMware console and check the settings for the VM's network adapters, I find that the Connected box is not checked, even though Connect at power on is checked. Is this a known issue? -- Michael Jinks :: mji...@uchicago.edu University of Chicago IT Services