On 02/04/2020 12:42 PM, David C. Rankin wrote:

On 02/04/2020 04:18 AM, David C. Rankin wrote:

...

lAll,

After update to 5.5 kernel and rebuild of virtualbox from the 5.2.36 and the latest testcase build for 5.2.37 (all kernel modules build and load fine)

https://www.virtualbox.org/download/testcase/VirtualBox-5.2.37-135942-Linux_...

Arch guests fail to run on Arch host. network device enp0s3 not found, udev kernel rules fail, and the boot never complete.

All fine with the 5.4 kernel. I've posted to the vbox-users list, but if anyone here has additional information I would appreciate it.

This explains Virtualbox does not support the 5.5 kernel -- no fix yet:

Ticket #19145 (accepted defect) Opened 8 weeks ago linux: kernel 5.5-rc1/rc2 - we need changes https://www.virtualbox.org/ticket/19145

OK, Need help from the smart folks. Downgrading the kernel/kernel-headers to 5.4.15 and downgrading virtualbox to the version working on 5.14 [testcase 5.2.35 (135458)] does not allow guests to start as they did before the last update to linux 5.5. Starting the guest headless - reports it started. However, when I open the guest via-rdesktop monitor the boot, but the boot process would not complete and could not create the network interface enp0s3, the udev kernel device manager will not start, the boot process cannot flush the journal to persistent storage (??) and boot cannot create volatile files and directories (??) All modules build for each version of virtualbox. (5.2.35, 5.2.36 and testcase 5.2.37 (135942)). I have tried on the downgraded 5.4.15 and LTS. There is no combination that will start the virtualbox guests as before. There must be another package that was updated that comes into play (python?). Is anyone else experiencing problems with Virtualbox and has anyone else seen similar behavior? Any workarounds. I have a number of virtualbox guests I use. -- David C. Rankin, J.D.,P.E.

On 02/05/2020 09:31 AM, Erich Eckner wrote:

Is there a specific reason, that prevents you from using virtualbox 6.1.2 from [community] - am I missing something here?

Yes, Unfortunately Win7 guests don't play well with 6.x over the network when started headless and I need to extension pack. (I should try again, I don't know if the problem was fixed in the past few 6.x releases after I reported it early on). That's why I have continued to build the 5.2 branch for AUR under virtualbox-bin-5. Oracle had indicated late 2020 would be the last 5.2 release, but they are still continuing their support for that branch. (I suspect I'm not the only one that stumbled into the network deploy issue with 6.x) The only reason I don't know more testing with 6.x is it requires a backup of the VM's config as 6.x makes changes that have to be undone to downgrade back to 5.2. (but at this point, it's worth giving it another shot) -- David C. Rankin, J.D.,P.E.

On 02/05/2020 10:01 AM, Ralph Corderoy wrote:

Hi David,

...

There must be another package that was updated that comes into play (python?). How about restoring all packages to an earlier date? https://wiki.archlinux.org/index.php/Arch_Linux_Archive#How_to_restore_all_p...

The last update on the Archlinux guest was 1/27/20 (5.4.15 kernel), I downgraded to 1/21/20 (5.4.13 kernel) and the problem remains. I know there is something totally broken as it takes the guest over 600 seconds (10 Minutes) to boot to a minimal console login. On 1/27/20 it booted fine when I updated to the 5.4.15 kernel. The only packages that got downgraded were: ca-certificates ca-certificates-mozilla ca-certificates-utils dbus 1.12.16-3 1.12.16-5 htop hwinfo imagemagick intel-tbb 2020.1-1 2020.0-1 libsecret 0.20.0-1 0.19.1-3 libutil-linux linux 5.4.13 5.4.15 linux-headers 5.4.13 5.4.15 nss p11-kit perl-timedate python python-six python2-cairo unace util-linux vim vim-runtime whois xfsprogs (I only typed the versions from the screenshot for the packages that seemed like they might be relevant) It's almost like some module or config was updated that broke something on the guest, so now it doesn't matter whether I downgrade or not, the problem remains. Most amazing frustrating problem I run into in a while. I was able to grad a dmesg output and the last 1000 lines of the packman.log, but I'm stumped. -- David C. Rankin, J.D.,P.E.

On 02/14/2020 03:00 AM, David C. Rankin wrote:

A bug (https://www.virtualbox.org/ticket/19311) was opened with Oracle concerning this issue that has broken the AUR package https://aur.archlinux.org/packages/virtualbox-bin-5/

Just a follow-up advising that the AUR virtualbox-bin-5 package updated to 5.2.38 is fine and all bugs related to apparent issues with the Linux 5.4 to 5.5 update are resolved. (the last issue, high io_wait between host and guest was due to high waits on an underlying disk in the array providing guest storage - no errors in journal and mdadm reported all disks in the array good, but /dev/sdc was beginning to die -- once a scrub of the array started, the estimated 26 days to complete the scrub on a 3T array was a dead giveaway...) -- David C. Rankin, J.D.,P.E.