Recent content by efeu

Hi, (Basicaly you have to get RDMA working first. Look here: https://enterprise-support.nvidia.com/s/article/howto-configure-nfs-over-rdma--roce-x) it took me a while to find a way how proxmox could use NFSoRDMA. First I tried to change the storage options in /etc/pve/storage.cfg from nfs...

Yeah, by reseting the server there are plenty of options.

How would you get root access? You would still need the root credentials to log in to console? Or do you mean by shuting down server and get root access via chroot?

Sure, (but) this is already built into proxmox. CPU pinning is not

While running low latency tasks in VMs, you may want to pin your vcores to physical cores to not loose caches and avoid moving tasks to cores which are running on lower frequency first when moving to them. The existing pve_helper scripts managed to pin the qemu CPU threads to a "CPU pool"...

Hi, I added a cache drive to my pool which has secondarycache=none attribute. pool: spinning state: ONLINE scan: scrub repaired 0B in 04:40:29 with 0 errors on Sun Oct 9 05:04:31 2022 config: NAME STATE READ WRITE CKSUM spinning...

Well after an unwanted switch power down, I dont see any path drops anymore. So I guess it's working as it should.

I currently testing this. I got pNFS working (for a while) by doing it this way: modprobe nfs_layout_nfsv41_files modprobe nfs_layout_flexfiles mount x.x.1.x:/some/path /some/other/path -o vers=4.1,minorversion=1,max_connect=16 mount x.x.2.x:/some/path /some/other/path -o...

The solution is to set echo 1 > /proc/sys/net/ipv6/conf/eth0/accept_ra (or 2) inside the container 0 Do not accept Router Advertisements. 1 Accept Router Advertisements if forwarding is disabled. 2 Overrule forwarding behaviour. Accept Router Advertisements even if forwarding is...

Hi, I'm running a debian10 lxc container and want to set the ipv6 token. But when running ip token set ::xxx dev eth0 inside the container it returns Error: ipv6: Router advertisement is disabled on device. But actual it does get an IP from SLAAC. inet6 xxx/64 scope global dynamic mngtmpaddr...

Then after about 3 hours of build, it aborts with the following error(s) .... # Autogenerate blacklist for watchdog devices (see README) install -m 0755 -d debian/pve-kernel-5.4.41-1-pve/lib/modprobe.d ls debian/pve-kernel-5.4.41-1-pve/lib/modules/5.4.41-1-pve/kernel/drivers/watchdog/ >...

How do you build pve kernel in parallel? The -j flag on make ist completly ignored and building the kernel on a low clocked 32core epyc takes like forever (but well, could use 64 threads instead, which should speedup build times enormously)

Build mainline kernel 5.6 with standard ubuntu patches and zfs 0.8.3 built in [-[ with 5.5/5.6 patches (until zfs 0.8.4 is out) ]-] apt install git build-essential kernel-package fakeroot libncurses5-dev libssl-dev ccache flex bison libelf-dev build-essential autoconf libtool gawk alien...

OK, I found it add: push @$cpuFlags , '-x2apic' if $cpu eq 'EPYC'; in sub get_cpu_options in /usr/share/perl5/PVE/QemuServer/CPUConfig.pm Sould be disabled by detecting AMD CPU

There seems to be an "older" bug in pve - when I enable kvm_amd avic, proxmox would add x2apic flag to and EPIC cpu (which is and intel cpu flag, so no wonder it cannot work) More information: https://bugzilla.redhat.com/show_bug.cgi?id=1675030 I could fix this by my own, so please link me to...