spirit's latest activity
-
spirit replied to the thread Cluster retransmit issues.About spanning-tree, you should really disable it on your physical switch port of for your proxmox nodes. a spanning tree convergence can happen on host reboot and broke the whole cluster for some second. you don't need change knet_mtu, it's... -
spirit replied to the thread Ceph - VM with high IO wait.Strange that you also have high memory pressure "PSI some memory". do you have enable numa option on the vm ? you can also look at host numa stat # apt install numactl # numstat and look if you don't have a lot of "numa_miss" vs "numa_hit"...
-
spirit replied to the thread Cluster retransmit issues.can you send corosync log of each node ? (journalctl -u corosync). is the nic for corosync link dedicated ? or do you have vm,storage,backup,...running on it too ? (no bandwidth saturation ? ) no spanning tree on the network ? do you use...
-
spirit replied to the thread Interest in VPP (Vector Packet Processing) as a dataplane option for Proxmox.firewall=1 is create a linux bridge fwbr interface, so it's not usable here. Implementing the whole firewall code in vpp is another thing. (and currently proxmox code don't have any easy plugin code to implement different firewall)
-
spirit replied to the thread How to disable the bridge level firewall in PVE 9.1.1.So, why do you want to use pve-firewall ? (I mean, disable pve-firewall service or uncheck firewall checkbox on the vms) I'll try to look at proxmox-firewall code, but it shouldn't be needed
-
spirit reacted to fba's post in the thread Interest in VPP (Vector Packet Processing) as a dataplane option for Proxmox with
Like.
As far as I understand the architecture it is for one part the single packet processing for an entire pipeline vs. handling batches of packets (aka vectors) at each step of the pipeline. The latter one saves time when loading the cpu instructions... -
spirit replied to the thread Interest in VPP (Vector Packet Processing) as a dataplane option for Proxmox.maybe the best way is to ask to the dev mailing list pve-devel@lists.proxmox.com. (I'm pretty sure that some users could be interested for routers vm appliance) the basic dev doc for patch submission is here...
-
spirit replied to the thread How to disable the bridge level firewall in PVE 9.1.1.This is needed if you want to use iptables (used by pve-firewall) to have ip rules at bridge level. why do you want to disable them ? Alternatively, they shouldn't be needed by the new nftables based firewall (proxmox-firewall services), as...
-
spirit replied to the thread Interest in VPP (Vector Packet Processing) as a dataplane option for Proxmox.Hi, I don't remember, but does it work with virtio-net ? I thinked it need vhost-user-net (that is not implement in proxmox currently )
-
spirit replied to the thread rp_filter changes in PVE 9.mmm, this seem to be a change in debian13: https://www.debian.org/releases/trixie/release-notes/issues.html#etc-sysctl-conf-is-no-longer-honored In Debian 13, systemd-sysctl no longer reads <span>/etc/sysctl.conf</span>. The package...
-
spirit replied to the thread NICs inoperative in ProxMox.say thanks you to systemd. the nic naming is based of pci slot ordering. sometimes, when adding a pcie devices (or nvme drive), the internal order can change. (depend of the motherboard). pve9 have a new feature to add a statc name "nicX" based...
-
spirit replied to the thread A large number of dropped packets.do you use bonding on your proxmox node ? if yes, which mode ? dropped traffic could be multicast or unicast flood with destination ip is not the ip of our vm. (check also if mac address ageing timeout not too low on your physical switch)
-
spirit replied to the thread A large number of dropped packets.always use virtio. e1000 don't have any acceleration
-
spirit replied to the thread Ceph rbd du shows usage 2-4x higher than inside VM.I known that ext4 had problem with discard in the past (not about fragmentation, but discard not always working). Personally, I'm using xfs in production, and I never had this problem (on 4000 vms)
-
spirit replied to the thread Ceph rbd du shows usage 2-4x higher than inside VM.do you have any snapshot in theses vms ? (because triming on snapshot will take more space instead reduce the space)
-
spirit replied to the thread SDN overlay network in routed mesh setup.they are an option on the zone: "exit nodes local routing"
-
spirit replied to the thread ZFS mirror on 2x Crucial T705 (PCIe 5.0) causing txg_sync hangs under write load – no NVMe errors in dmesg.(small reminder: don't use zfs on consumer ssd/nvme . they can't handle a lot of fsync because they don't have a PLP/powercapacitor), and zfs do a lot of sync. It's really like 200~1000 iops max with this kind of drive.
-
spirit replied to the thread Network Optimization for High-Volume UDP Traffic in PVE.yes, I was thinking exactly the same
-
spirit replied to the thread Network Optimization for High-Volume UDP Traffic in PVE.this is normal, don't use vmxnet3 or e1000, they are full software emulation. you need to use virtio which use vhost-net offloading on pvehost. your cpu is quite old, and it's possible that spectre/meltdown/.... mitigation impact performance...
-
spirit replied to the thread Network Optimization for High-Volume UDP Traffic in PVE.250566pps is quite low, I mean , you should reach 1~2mpps for any each packetsize. I remember to reach easily 7~9gbit with 1core/thread with standard 1500mtu. (with epyc v3 3,5ghz and cpu forced to max frequency)