TomE

I have been experiencing an issue with virtiofs mounts hanging on my debian guests, changing XML virtiofsd config hasn't helped. There's a similar issue report at https://gitlab.com/virtio-fs/virtiofsd/-/work_items/133 which references a fix in QEMU versions 10.2.0 or higher. It doesn't look like it's been added to QEMU 10.1.2. Is there a reason for why this specific version of QEMU was picked?

I managed to fix this by removing the "cache mode" line inside the virtiofsd args: <filesystem type='mount' accessmode='passthrough'> <driver type='virtiofs' queue='1024'/> <binary path='/usr/libexec/virtiofsd' xattr='on'> <sandbox mode='chroot'/> </binary> <source dir='/mnt/user/music/'/> <target dir='music'/> <alias name='fs2'/> <address type='pci' domain='0x0000' bus='0x05' slot='0x00' function='0x0'/> </filesystem> I believe this changes the cache mode from "always" to "auto" - it may compromise performance slightly. Libvirt only lists two options "none" and "always" in the XML so by removing it I assume it's using the "auto" setting (default in the virtiofsd manual).

Same behavior on Unraid 7.1.2, files moved over SMB share not showing

OK, I just ran the experiment again. This time I tried on Unraid 6.10.0-rc5 and ran a linux server (Ubuntu Server 20.04) and a windows server (Windows Server 2022). On both I used Temurin OpenJDK "jdk-17.0.3+7". Ran a timings report on both and found that the linux server was actually faster (as it should be). Quite pleased by this result, starting to think it is the newer version of libvirt used in 6.10 so will probably upgrade as soon as a stable release is available. Could possibly be due to a different jdk in earlier testing? But anyway here's the timing costs of windows vs linux: timingcost: 150 name: Windows Server 2022 version: 10.0 jvmversion: 17.0.3 arch: amd64 maxmem: 12884901888 cpu: 12 runtime: 371191 flags: -Xms12G -Xmx12G -XX:+UseG1GC -XX:+ParallelRefProcEnabled -XX:MaxGCPauseMillis=200 -XX:+UnlockExperimentalVMOptions -XX:+DisableExplicitGC -XX:+AlwaysPreTouch -XX:G1HeapWastePercent=5 -XX:G1MixedGCCountTarget=4 -XX:G1MixedGCLiveThresholdPercent=90 -XX:G1RSetUpdatingPauseTimePercent=5 -XX:SurvivorRatio=32 -XX:+PerfDisableSharedMem -XX:MaxTenuringThreshold=1 -XX:G1NewSizePercent=30 -XX:G1MaxNewSizePercent=40 -XX:G1HeapRegionSize=8M -XX:G1ReservePercent=20 -XX:InitiatingHeapOccupancyPercent=15 -Dusing.aikars.flags=https://mcflags.emc.gs -Daikars.new.flags=true timingcost: 128 name: Linux version: 5.15.0-27-generic jvmversion: 17.0.3 arch: amd64 maxmem: 12884901888 cpu: 12 runtime: 205833 flags: -Xms12G -Xmx12G -XX:+UseG1GC -XX:+ParallelRefProcEnabled -XX:MaxGCPauseMillis=200 -XX:+UnlockExperimentalVMOptions -XX:+DisableExplicitGC -XX:+AlwaysPreTouch -XX:G1HeapWastePercent=5 -XX:G1MixedGCCountTarget=4 -XX:G1MixedGCLiveThresholdPercent=90 -XX:G1RSetUpdatingPauseTimePercent=5 -XX:SurvivorRatio=32 -XX:+PerfDisableSharedMem -XX:MaxTenuringThreshold=1 -XX:G1NewSizePercent=30 -XX:G1MaxNewSizePercent=40 -XX:G1HeapRegionSize=8M -XX:G1ReservePercent=20 -XX:InitiatingHeapOccupancyPercent=15 -Dusing.aikars.flags=https://mcflags.emc.gs -Daikars.new.flags=true

Hi, I'm having similar issues (running a Ryzen 5 3rd gen) with a linux server VM. At the moment i'm not sure if it is a java specific (JVM related) thing or if it is a virtualization problem as I have only tested the minecraft server on both platforms. Running a windows server VM (as its lighter weight than full windows install) with the correct drivers (virtio installed) I was able to get good performance which was close to the CPU's actual speed when running a benchmark (as one would hope). I also tried running a minecraft server on this windows VM and the performance was great, TPS good. After switching back to Linux I was greeted with terrible performance for the minecraft server, which is annoying because it uses far less idle resources on Linux which is why i prefer it. I still need to do more testing so I need a good multi-platform benchmark software to properly compare them. I think upgrading to UNRAID version 6.10.0 may fix the issue due to it having an updated virtualization software and if that doesn't fix it then I could try get a baseline through Proxmox or another virtualization platform (proxmox is closest to unraid) and maybe bring it up with the devs. But I need to do more testing. Tom.

What backup software did you end up using?

Hello, I am having issues with the network slowing down while torrenting. I am using the Nord OpenVPN profiles and have tried two with the same issues. The torrents become really slow then stop. When looking on my router, the packet loss increases to around 70% and latency also increases and the speeds for everyone on the network drop. In our old setup, we had the same issue until we changed it to the NordLynx protocol and all the issues went away, although I cannot do this on the docker. I have a feeling that the ISP might be throttling. I would like to know if there is any other way to get this to work. Many Thanks, Tom.