Description of problem: The VM provisioned for RHEL 8.4 in Azure will fail to do a first reboot if it's done from Azure portal. Post troubleshooting and checking all the things as kdump,WaLinuxAgent customer found the shim package is responsible for this. How reproducible: Always post provisioning the VM and rebooting it from portal with a restart switch/button. Steps to Reproduce: [1] Get the Azure Cli and provision a VM as, $az login $az account set --subscription <your sub id> $az group create -l <location (I prefer westus2)> -n <Resource Group name> $az vm create --name <VM Name> --resource-group <RG Name that is create by above command> --location <westus2> --image "RedHat:RHEL:84-gen2:8.4.2021053102" --size Standard_D2s_v3 --admin-username <username> --admin-password <password> Secondly, you can go through this link for reference : https://docs.microsoft.com/en-us/azure/virtual-machines/workloads/redhat/redhat-images [2] Once done Open the Azure portal and click the restart button. The VM will notify about rebooting but will never come up. [3] Upon hard - reset or restart as Stop and Start the VM will start and there after the "restart" will work as expected. *** The issue occurs only at first reboot done using restart button" *** Version-Release number of selected component (if applicable): shim-x64-15.4-2.el8_1.x86_64 Actual results: System fails to reboot with nothing printed to console as suspect, no screenshot as there is no direct console and so far I see it fails to return the boot status to arm signal sent by the webcode to hypervisor. Hard reset boots it up and next "restarts" work fine so shims seems to be adapting something. Expected results: Shall not fail to reboot at first reboot from Portal call. Additional info: Customer did check as, We have RHEL 8.4 shim-x64-15.4-2.el8_1.x86_64 RHEL 8.3 shim-x64-15-16.el8.x86_64 So even if we update the 8.3 release with shim-x64-15.4-2.el8_1.x86_64 it will fail to boot once. So I suspect the arm calls from portal are failing as while checking this more I have checked journal and can see a kill sent as normal shutdown. Limited to Hyper-V env so can't have direct serial console logs and I have asked to share them, will forword if get available. Downgrading the shim to shim-x64-15-16.el8.x86_64 on RHEL 8.4 image makes the issue go away!
At present I cannot get to a point where I can even try to reproduce this issue. While we work that out, anyone who *can* reproduce the issue, please: 1) install shim-x64-15-16.el8.x86_64 2) from linux, run "mokutil --set-verbosity true" as root 3) if possible, copy \EFI\redhat to \EFI\test , put the shimx64.efi from shim-x64-15.4-2.el8_1.x86_64 there, and set it up as a second boot entry in the system firmware 4) capture the console log while restarting with this new boot entry selected
(In reply to Peter Jones from comment #7) > At present I cannot get to a point where I can even try to reproduce this > issue. While we work that out, anyone who *can* reproduce the issue, please: > > 1) install shim-x64-15-16.el8.x86_64 > 2) from linux, run "mokutil --set-verbosity true" as root > 3) if possible, copy \EFI\redhat to \EFI\test , put the shimx64.efi from > shim-x64-15.4-2.el8_1.x86_64 there, and set it up as a second boot entry in > the system firmware > 4) capture the console log while restarting with this new boot entry selected Hi Peter, If I run "mokutil --set-verbosity true" and then restart the VM, the VM shows a "Secure Boot not enabled" blue screen, and because Azure only have serial console, it cannot get passed. Besides, Azure cannot select boot order, so I have to set the new entry as the first boot order.
Hi pjones, Setting verbosity indeed drops you to a OK prompt, so that's problematic since console is not available. Would it be possible to provide a shim-x64 package that doesn't show that OK prompt?