VMware boot disk

Support
, , ,
#1

Problem:
I’m looking at using the bootdisk plugin to automatically provision VMware hosts. It works fine in BIOS mode, but I’m hoping to use EFI. I’m thinking this doesn’t currently work because the provisioning process gets forced to use the minimal host image, but to be usable on EFI it would need the full host image. Is there a way as a user to force this automated workflow to use the full host image as opposed to the minimal one?

Foreman and Proxy versions:
2.1

Distribution and version:
CentOS 7.8

#2

Full host bootdisk is EFI compatible.

1 Like
#3

Right. But can I use it in the automated provisioning workflow? I don’t see an option to choose which image, it just seems to use the minimal host image no matter what

#4

What’s minimal host image? Can you paste some screenshots or CLI commands with expectations how things should work?

#5

Absolutely. So I go through the workflow to create a host on my VMware compute resource:

image
image1227×682 28 KB

Then I switch to the OS tab and select boot disk:

image
image1455×700 36.2 KB

I’ll also show EFI selected in the VM tab:

image
image1047×249 10.6 KB

Then I go ahead and click submit and get a host in VMware, but see this:

image

I believe this is because it’s using a per-host image as opposed to a full host image (sorry if the nomenclature I used earlier was wrong). I’m unclear if there’s a way to force it to use a full host image.

#6

Try this patch (untested):

#7

Thanks for looking into this! I’m not sure I know the proper way to apply this patch to an existing foreman installation. I tried anyway, ended up with this error:

image
image1588×148 10.3 KB

#8

Revert the patch, apply again (I rebased an overlook).

#9

New error this time:

image
image1560×79 4.73 KB

#10

I don’t have VMWare so I can’t reproduce it here, can you revert and apply again? This time I added a logging message to find out what kind of object is this called on. Ruby can be sometimes terribly confusing when objects are extended and the context is not clear. Send me the info log line that reads ISO IMAGE OBJECT.

#11

Here you go:

2020-10-14T05:15:18 [I|app|0e239341] ISO IMAGE OBJECT: #<Host::Managed id: nil, name: “felix-winfrey.jeff.dev”, last_compile: nil, last_report: nil, updated_at: nil, created_at: nil, root_pass: [FILTERED], architecture_id: 1, operatingsystem_id: 1, environment_id: nil, ptable_id: 116, medium_id: 1, build: true, comment: “”, disk: “”, installed_at: nil, model_id: nil, hostgroup_id: 1, owner_id: 4, owner_type: “User”, enabled: true, puppet_ca_proxy_id: nil, managed: true, use_image: nil, image_file: nil, uuid: “501ffe9f-a53d-3a56-d59e-939f763e1f54”, compute_resource_id: 1, puppet_proxy_id: nil, certname: nil, image_id: nil, organization_id: 1, location_id: 2, type: “Host::Managed”, otp: nil, realm_id: nil, compute_profile_id: 1, provision_method: “bootdisk”, grub_pass: “$5$T3Ho9bookfZCZedd$g7aXK9eQAxrVSmGe34HW4lQ141lzYF…”, global_status: 0, lookup_value_matcher: [FILTERED], pxe_loader: “None”, initiated_at: nil, build_errors: nil>

#12

Do you still see the same error? Then I need to see the backtrace, because this comes probably from different code than I thought.

#13

I do see the same error. Was something supposed to have changed since yesterday? Code looks the same to me. Thanks

#14

No changes I just need to see ruby stacktrace to find out the code which actually fails with this error.

#15

I’m not great with ruby, so not really sure how to provide that stacktrace

#16

When you encounter the error you showed on the screenshot, go to production.log and there should be an error message with many .rb files and line numbers. Pastebin me that.

#17

Here you go: https://pastebin.com/TRS4xM4U

#18

Damn, the stacktrace is swallowed. Can you try again now with debug logging turned on? Foreman core should log the backtrace for rolled transactions in INFO log, I wonder why this does not work.

Just before trying again do one additional patch:

diff --git a/app/models/concerns/foreman_bootdisk/orchestration/compute.rb b/app/models/concerns/foreman_bootdisk/orchestration/compute.rb
index dfed569..5809bf4 100644
--- a/app/models/concerns/foreman_bootdisk/orchestration/compute.rb
+++ b/app/models/concerns/foreman_bootdisk/orchestration/compute.rb
@@ -71,6 +71,7 @@ module ForemanBootdisk
       def setGenerateIsoImage
         bootdisk_generate_iso_image
       rescue StandardError => e
+        Rails.logger.info "BT (#{exception.class}): #{exception.message}\n" + backtrace.join("\n")
         failure format(_('Failed to generate ISO image for instance %{name}: %{message}'), name: name, message: e.message), e
       end
#19

Sure, here you go: https://pastebin.com/wWUTAiDt

#20

I made a typo when I was doing the last patch :slight_smile: It should have been:

diff --git a/app/models/concerns/foreman_bootdisk/orchestration/compute.rb b/app/models/concerns/foreman_bootdisk/orchestration/compute.rb
index dfed569..058a597 100644
--- a/app/models/concerns/foreman_bootdisk/orchestration/compute.rb
+++ b/app/models/concerns/foreman_bootdisk/orchestration/compute.rb
@@ -71,6 +71,7 @@ module ForemanBootdisk
       def setGenerateIsoImage
         bootdisk_generate_iso_image
       rescue StandardError => e
+        Rails.logger.info "BT (#{e.class}): #{e.message}\n" + backtrace.join("\n")
         failure format(_('Failed to generate ISO image for instance %{name}: %{message}'), name: name, message: e.message), e
       end

It’s exception vs e :stuck_out_tongue: