Is there a version of PVE that uses systemd-boot? Everytime I installed PVE it only ever came with the GRUB bootloader
[SOLVED] Network Issues after updating
- Thread starter Kaoplo
- Start date
Good point! I forgot about that! Thanks for pointing it out.
To update - disabling intel_iommu and monitoring the host overnight I haven't seen any issues with the network. Looks like that is the solution so I'll be applying it to my whole cluster.
I just got my rig setted up, added a consolebreak command in /etc/default/grub, reboot it and some how got my rig unreachable from other devices. Been looking up several hours as my network config seems fine and no problem to be found.This ultimately fixed it. Thanks a lot man.Yeah, that seems like it so far
If you edit/etc/default/gruband addintel_iommu=offto theGRUB_CMDLINE_LINUX_DEFAULTproperty followed by runningupdate-gruband a reboot it seems like that fixes it.
2 hours so far with no issues. I'll report back in the morning to be sure.
NOT SOLVED...
I've been battling this for almost a year now and nothing worked. I did turn intel_iommu=off and on too.
Eight months ago I had 3 Open vSwitch bridges in use, two exclusively for a pfSense VM and another for ProxMOX and other VM's. When I ran on PVE 7.x it worked pretty much smoothly although I had issues with a dead WAN PPPoE NIC every now and then. Basically it worked for several weeks until the WAN went down.
Updated to PVE 8.x and that made things worse. Over the course of 7 months I got these well known timeouts. At first only every 2 up to 4 weeks. Tested all kinds of situations with intel_iommu=off and on. Some people suggested that this issue may be triggered by high network traffic load, so I've tested this with all kinds of limiters - no difference what so ever.
Currently I run two NIC's in PCI pass through and one still as Open Switch bridge for ProxMOX and one VM. Again... only the two passed through NIC's work flawlessly now and the other one has an almost daily timeout of 5 minutes. And that port has very low network activity.
The only other tweak that's left is changing the NIC firmware, but I don't know if that will work on a Mac mini 2014 and Apple Thunderbolt Ethernet NIC's.
https://forum.proxmox.com/threads/broadcom-nics-down-after-pve-8-2-kernel-6-8.146185/page-3
https://www.thomas-krenn.com/de/wiki/Installation_NICCLI_unter_Proxmox_VE
I've been battling this for almost a year now and nothing worked. I did turn intel_iommu=off and on too.
Eight months ago I had 3 Open vSwitch bridges in use, two exclusively for a pfSense VM and another for ProxMOX and other VM's. When I ran on PVE 7.x it worked pretty much smoothly although I had issues with a dead WAN PPPoE NIC every now and then. Basically it worked for several weeks until the WAN went down.
Updated to PVE 8.x and that made things worse. Over the course of 7 months I got these well known timeouts. At first only every 2 up to 4 weeks. Tested all kinds of situations with intel_iommu=off and on. Some people suggested that this issue may be triggered by high network traffic load, so I've tested this with all kinds of limiters - no difference what so ever.
Currently I run two NIC's in PCI pass through and one still as Open Switch bridge for ProxMOX and one VM. Again... only the two passed through NIC's work flawlessly now and the other one has an almost daily timeout of 5 minutes. And that port has very low network activity.
Code:
pve kernel: tg3 0000:03:00.0 enp3s0f0: NETDEV WATCHDOG: CPU: 1: transmit queue 0 timed out 6720 ms
pve kernel: tg3 0000:03:00.0 enp3s0f0: transmit timed out, resetting
pve kernel: tg3 0000:03:00.0 enp3s0f0: 0x00000000: 0x168614e4, 0x00100406, 0x02000001, 0x00800040The only other tweak that's left is changing the NIC firmware, but I don't know if that will work on a Mac mini 2014 and Apple Thunderbolt Ethernet NIC's.
https://forum.proxmox.com/threads/broadcom-nics-down-after-pve-8-2-kernel-6-8.146185/page-3
https://www.thomas-krenn.com/de/wiki/Installation_NICCLI_unter_Proxmox_VE
Try to use "intel_iommu=on iommu=pt" instead of "intel_iommu=off."
This solved my problem.