Booted engine host and centos hypervisor host. both at 4.2.3 GA. Upgraded engine to 4.2.4 GA release. Host is marked as non operational at boot with available upgrade. Requested upgrade with reboot after upgrade. At reboot kdump image is not found and being built. At the end of kdump image creation host is stuck asking for root password. kdump image should be generated as part of the upgrade process.
Marked as severity urgent since this require people to physically go to the host and manually reboot.
(In reply to Sandro Bonazzola from comment #0) > Booted engine host and centos hypervisor host. both at 4.2.3 GA. > Upgraded engine to 4.2.4 GA release. > Host is marked as non operational at boot with available upgrade. > Requested upgrade with reboot after upgrade. > At reboot kdump image is not found and being built. At the end of kdump > image creation host is stuck asking for root password. > kdump image should be generated as part of the upgrade process. How is this connected to the oVirt host upgrade flow? We are only executing "yum update" and when finished successfully, we are rebooting the host using SSH connection and invoking "systemctl reboot". Kdump image regeneration is automatic and it should happen after each kernel upgrade. If Kdump is asking for root password after reboot, shouldn't this be filed against kdump?
(In reply to Martin Perina from comment #5) > How is this connected to the oVirt host upgrade flow? We are only executing > "yum update" and when finished successfully, we are rebooting the host using > SSH connection and invoking "systemctl reboot". Kdump image regeneration is > automatic and it should happen after each kernel upgrade. If Kdump is asking > for root password after reboot, shouldn't this be filed against kdump? I'm not sure if the issue is caused by a bug in kexec-tool package or if it's caused by the host kdump configuration we do on host deployment. If you think this is not a bug in ovirt upgrade flow, I would suggest to open a bug on kexec-tools blocking this one and either add a warning before starting upgrade from the engine or not allowing upgrading if kdump is enabled on the target host.
Ravi, could you please try to reproduce it (we probably need to be able to upgrade both kernel and kexec-tools packages during host upgrade)? If it's really reproducable even using plain yum update, please file a bug on kexec-tools
I am unable to reproduce the issue. Both "yum update kernel kexec-tools" and update via Host Upgrade from Webadmin work as they should. Is this reproducible 100% of the time?
Closing as worksforme, feel free to reopen if reproduced and provide clear reproducing steps