Questions and Answers : Unix/Linux : Can't run Rosetta tasks using VirtualBox afterupgrade to Ubuntu 22.04
Author | Message |
---|---|
Penumbra Du Fromage Send message Joined: 1 Aug 06 Posts: 4 Credit: 1,280,295 RAC: 1,308 |
I'm x-posting this from the BOINC forums, as there's no reply there: When running a task for Rosetta@Home - BOINC Can't access the VirtualBox VM's properly. My system is Ubuntu 22.04.1 running VirtualBox Version 6.1.34_Ubuntu r150636 In trying to solve this problem, I've tried a couple of versions of BOINC now: Ubuntu's package manager installed: 7.18.1 - This wouldn't even run.. Illegal instruction error (core dumped) - This triggered me to try another version Costamagna Gianfranco-ubuntu-boinc: 7.20.2 - Runs but all tasks using VirtualBox error out with: 2022-08-16 16:40:02 (180543): Error in host info for VM: -1041038848 - If I copy/paste the exact command listed 'VBoxManage -q list hostinfo' - I see this: → VBoxManage -q list hostinfo Build directly from Source: 7.21.0 x86_64-pc-linux-gnu - SAME ERRORS AS ABOVE VirtualBox is installed, the command works as myself, the user 'boinc' (which runs the service) or as root. What it doesn't seem to do it run when BOINC tries to launch a task using it. Before I downgrade - does ANYONE have an idea on why this is happening, or next steps for me to try to figure it out myself? I'm quite linux savvy, and handy with a compiler - but I don't yet see where this happens. |
Jim1348 Send message Joined: 19 Jan 06 Posts: 881 Credit: 52,257,545 RAC: 0 |
My system is Ubuntu 22.04.1 running VirtualBox Version 6.1.34_Ubuntu r150636 It is due to the new security setting in BOINC 7.18.1 and above. https://quchempedia.univ-angers.fr/athome/forum_thread.php?id=166#1644 You will need to go back to 7.16.6. I don't know if that is possible in Ubuntu 22.04.1. That is why I stay with 20.04.4. |
Penumbra Du Fromage Send message Joined: 1 Aug 06 Posts: 4 Credit: 1,280,295 RAC: 1,308 |
It is due to the new security setting in BOINC 7.18.1 and above. I tried the workaround on that thread to no avail. I will see about backing myself down to 7.16.6 if possible.. I can build from source too - so if I can trace the patches that caused this from that thread, I may be able to produce a new version which works. There's something really wonky going on in Ubuntu 22 for sure. Even with the BOINC client service running, but all tasks suspended - I get a massive spam in my logs about CRT disconnects: Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): CRT-0: disconnected Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): CRT-0: 400.0 MHz maximum pixel clock Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): HP 27es (DFP-0): connected Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): HP 27es (DFP-0): Internal TMDS Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): HP 27es (DFP-0): 340.0 MHz maximum pixel clock Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): VIZ VL320M (DFP-1): connected Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): VIZ VL320M (DFP-1): Internal TMDS Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): VIZ VL320M (DFP-1): 340.0 MHz maximum pixel clock Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): CRT-0: disconnected Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): CRT-0: 400.0 MHz maximum pixel clock Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): HP 27es (DFP-0): connected Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): HP 27es (DFP-0): Internal TMDS Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): HP 27es (DFP-0): 340.0 MHz maximum pixel clock Aug 26 09:40:52 Traal /usr/libexec/gdm-x-session[9557]: (--) NVIDIA(GPU-0): (Repeat ad infinitum) I have 2 monitors (VIZ VLM320M HDMI TV, and an HP 27es) -- the CRT-0 is not something I recognize as one of my connections, so I'm assuming this is a BOINC GPU session going haywire. This spam results in my monitor's blacking out every second or so for a few seconds. If I stop boinc-client service, the spam ends and the monitor regains fidelity. This may be unrelated, but it's confounding my investigations. |
Penumbra Du Fromage Send message Joined: 1 Aug 06 Posts: 4 Credit: 1,280,295 RAC: 1,308 |
I decided to pull the rip-cord, and I downloaded the 7.16.6 package versions for Ubuntu 20.0.4 (libboinc7 + boinc-client). The CRT GPU spam immediately stopped, and boinc now seems to function properly against normal work units. I'm still waiting for a virtualbox wu to download so I can see what happens. Will update as I see more.. |
Penumbra Du Fromage Send message Joined: 1 Aug 06 Posts: 4 Credit: 1,280,295 RAC: 1,308 |
Virtualbox WU's are still broken .. Hooray for LTS versions of OS's....... |
Jim1348 Send message Joined: 19 Jan 06 Posts: 881 Credit: 52,257,545 RAC: 0 |
Virtualbox WU's are still broken .. Hooray for LTS versions of OS's....... Against my better judgement, I tried 22.04 on their representation that you could access it remotely from a Windows machine using Remote Desktop. Sure enough, it did not work at all. I had to go back to X11 VNC, and even that didn't work, probably due to some additional security requirement buried somewhere. Each new version gets worse and worse. I only upgraded to 20.04 because of hardware compatibility, but never again if I don't have to. |
Questions and Answers :
Unix/Linux :
Can't run Rosetta tasks using VirtualBox afterupgrade to Ubuntu 22.04
©2024 University of Washington
https://www.bakerlab.org