Error I40E_AQ_RC_ENOSPC, forcing overflow promiscuous on PF
- Thread starter syfy323
- Start date
Code:
journalctl -b-1
Code:
dmesg | less
Code:
dmesg -Hl err -f kern... who is this troubled 0000:3:00.0 I wonder...
Code:
ls -l /sys/class/net/Ahhh the ole "VM-Private" IF w/ vmbr2 - super!
Code:
ethtool -i enp3s0f0np0version: 6.8.12-10-pve
firmware-version: 9.40 0x8000ecd6 0.0.0 <------- Lucky Me!
What is enabled/disabled??
Code:
ethtool -k enp3s0f1np1 | grep rx-vlanrx-vlan-filter: on [fixed]
rx-vlan-stag-hw-parse: off [fixed]
rx-vlan-stag-filter: off [fixed]
Code:
cat /etc/network/interfacesbridge-vlan-aware yes
bridge-vids 2-4094 <----- Hmm when did I set that?
&
Oh that Readme from Intel looks like a sexy bit of late night reading - Joy!
Once I wake up 5 times while reading, then get that hot new v2.27.8 driver slapped on and, get that config corrected, I'll report back. Thank you everyone from 2020 until today for sharing your findings <3
Yes! Sorry I got distracted..
I think the error is solved simply by changing the VLAN IDs setting (under each node's Networking --> vmbr*) from 2-4096 to whatever you're actually using.
So long as you don't have hundreds of VLANs...
If you do, you need to disable VLAN offloading to the NIC and instead use the CPU.
The 700 series NICs can only handle so many VLANs.
I gave it all an update
Bash:
apt update && apt full-upgrade -y && apt autoremove -y && apt clean
Bash:
wget https://github.com/intel/ethernet-linux-i40e/releases/download/v2.28.6/i40e-2.28.6.tar.gz
tar zxf i40e-2.28.6.tar.gz
rm i40e-2.28.6.tar.gz
cd i40e-2.28.6/src/
make install
update-initramfs -u
reboot
modinfo i40e... and then the firmware of the NIC
Bash:
apt install unzip -y
wget https://downloadmirror.intel.com/853788/700Series_NVMUpdatePackage_v9_54.zip
unzip 700Series_NVMUpdatePackage_v9_54.zip
tar zxf 700Series_NVMUpdatePackage_v9_54_Linux.tar.gz
rm 700Series_NVMUpdatePackage_v9_54_EFI.zip
rm 700Series_NVMUpdatePackage_v9_54_ESX.tar.gz
rm 700Series_NVMUpdatePackage_v9_54_FreeBSD.tar.gz
rm 700Series_NVMUpdatePackage_v9_54_Windows.zip
rm 700Series_NVMUpdatePackage_v9_54.zip
rm 700Series_NVMUpdatePackage_v9_54_Linux.tar.gz
cd 700Series/Linux_x64
chmod 755 nvmupdate64e
./nvmupdate64e
Select your NIC
01
Make a backup..
Y
<any key>
rebootEverything is all good now!
Hi Everyone,
I am also facing these issues and tried to find a solution but it seems to cause issues. Maybe someone can help.
As advices I updated to the latest version
This is my network card
This is my network config
Everything works as expected until I switch on the VLAN Awareness. (I need that in my network)
As I want to be flexible adding VLANs I have to use the whole range --> 2-4096 (Maybe I can leave >1500 out but same error)
So I understood that when using this I have to disable the VLAN Offloading and therefore added the Post up commands in the config.
I did that rebooted the system but then see this error in the log:
---> Goes up to 4096 ... Just don't want to mess up the Post here.
So I am really uncertain what to do know. Any ideas or suggestions?
Thanks Alex
I am also facing these issues and tried to find a solution but it seems to cause issues. Maybe someone can help.
As advices I updated to the latest version
Code:
root@pve1:~# uname -r
6.8.12-15-pveThis is my network card
Code:
01:00.0 Ethernet controller: Intel Corporation Ethernet Controller X710 for 10GbE SFP+ (rev 02)
01:00.1 Ethernet controller: Intel Corporation Ethernet Controller X710 for 10GbE SFP+ (rev 02)This is my network config
Code:
auto lo
iface lo inet loopback
iface eno1 inet manual
iface enxbe3af2b6059f inet manual
iface eno2 inet manual
iface enp1s0f0np0 inet manual
post-up /sbin/ethtool -K enp1s0f0np0 rxvlan off
post-up /sbin/ethtool -K enp1s0f0np0 rx-vlan-offload off
iface enp1s0f1np1 inet manual
post-up /sbin/ethtool -K enp1s0f1np1 rxvlan off
post-up /sbin/ethtool -K enp1s0f1np1 rx-vlan-offload off
#auto vmbr10
#iface vmbr10 inet static
# address 172.16.98.50/24
# gateway 172.16.98.254
# bridge-ports eno2
# bridge-stp off
# bridge-fd 0
auto vmbr0
iface vmbr0 inet manual
bridge-ports enp1s0f0np0
bridge-stp off
bridge-fd 0
bridge-vlan-aware yes
bridge-vids 2-4096
auto vmbr0.150
iface vmbr0.150 inet static
address 172.20.150.10/24
gateway 172.20.150.254Everything works as expected until I switch on the VLAN Awareness. (I need that in my network)
As I want to be flexible adding VLANs I have to use the whole range --> 2-4096 (Maybe I can leave >1500 out but same error)
So I understood that when using this I have to disable the VLAN Offloading and therefore added the Post up commands in the config.
I did that rebooted the system but then see this error in the log:
Code:
Sep 24 16:13:12 pve1 kernel: i40e 0000:01:00.0 enp1s0f0np0: entered allmulticast mode
Sep 24 16:13:12 pve1 networking[1206]: error: enp1s0f0np0: failed to set vid `{2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,---> Goes up to 4096 ... Just don't want to mess up the Post here.
So I am really uncertain what to do know. Any ideas or suggestions?
Thanks Alex
Last edited:
1. Your interface configuration looks very strange to me. I think that it needs to be re-done.
2. Flexibility to assign a new VLAN is good, but do you really need 4094 options? How many do you really have?
3. VLAN Hardware Offload should be set in the actual hardware NIC/Interface's settings, not the OS. Usually you want hardware offloading for throughput.... but these Intel 710 interfaces are not Beasts - they are cheap, that's why everything has them ;-) If you truly need thousands of VLANs you should buy a MUCH better NIC.
Hi,
thank you very much for your quick answer.
Can you please share which lines look strange? That would help a lot. If you mean the post-up then OK I already removed that but the rest should be OK
Of course not. With flexibility I mean sometimes I use higher number for VLANs etc. IT's a test Plattform. That's the only reason. All in All it is not more than 20-30. I can limit that if Hardware does not allow more for sure.
But good to find out some things here. In serious environments I fully agree - not a good idea
So if you have some more feedback with the config that would help.
Thanks
thank you very much for your quick answer.
Can you please share which lines look strange? That would help a lot. If you mean the post-up then OK I already removed that but the rest should be OK
No
Of course not. With flexibility I mean sometimes I use higher number for VLANs etc. IT's a test Plattform. That's the only reason. All in All it is not more than 20-30. I can limit that if Hardware does not allow more for sure.As I said - for testing
But good to find out some things here. In serious environments I fully agree - not a good idea So if you have some more feedback with the config that would help.
Thanks
Hello,
First off, I apologize - I hope I'm not necro-ing a thread from 4 months ago,
I also apologize if I missed anything while reading the discussion on this subject/bugfix...
I have follow-up questions on this issue...
------------
Background:
------------
I convinced my company to purchase multiple Proxmox subscriptions, which I am installing on old SuperMicro SYS-6018U-TR4T equipment, which seems to have this "NIC memory problem".
----
pve-manager/9.1.1/42db4a6cf33dac83 (running kernel: 6.17.2-1-pve)
81:00.0 Ethernet controller: Intel Corporation Ethernet Controller X710 for 10GbE SFP+ (rev 02)
driver: i40e
version: 6.17.2-1-pve
----
When I input 0-4094 for the Bridge-able VLAN IDs, and bring up the card, the linux console spams the Error message, and the whole system freezes for a good 5 minutes (except for the repeating messages). I can't even use CTRL-C or type at the root prompt.
I have stabilized the system by only specifying 50 VLANs.
We normally use VLAN numbers corresponding to internal project/customer numbers, so we use a random combination of very low VLAN IDs and very high VLAN IDs. I know it's possible to specify individual VIDs on the bridges as they change, but that's a huge pain.
---
Question #1:
My Boss is very mean, and demands to know why these Supermicro systems - which have been WORKING FINE WITH VMWARE for almost a decade - apparently have trouble doing basic VLAN routing with Proxmox, because of this problem. I know this isn't Proxmox's fault, but we've both been networking for a while, and never had to buy "beasty" network cards just to assign any VLAN IDs we wanted.
Question #2:
root@ci-proxmox04-cam-s:~# ethtool -k nic0 | grep vlan
rx-vlan-offload: on [fixed]
tx-vlan-offload: on [fixed]
This means I can't turn this feature off right?
Question #3:
Right now I am allowing only a range of 50 VLANs. How many can I allow before the memory becomes full? Is it trial and error?
Question #4:
Is there a danger of the card suddenly going crazy during production if I assign too many VLANs?
Question #5:
I do have access to the Enterprise Repo, but not on these particular older systems. Will the Enterprise Repo have updated drivers?
*** Thanks for any responses. This minor network glitch has really disrupted our rollout plans. ***
First off, I apologize - I hope I'm not necro-ing a thread from 4 months ago,
I also apologize if I missed anything while reading the discussion on this subject/bugfix...
I have follow-up questions on this issue...
------------
Background:
------------
I convinced my company to purchase multiple Proxmox subscriptions, which I am installing on old SuperMicro SYS-6018U-TR4T equipment, which seems to have this "NIC memory problem".
----
pve-manager/9.1.1/42db4a6cf33dac83 (running kernel: 6.17.2-1-pve)
81:00.0 Ethernet controller: Intel Corporation Ethernet Controller X710 for 10GbE SFP+ (rev 02)
driver: i40e
version: 6.17.2-1-pve
----
When I input 0-4094 for the Bridge-able VLAN IDs, and bring up the card, the linux console spams the Error message, and the whole system freezes for a good 5 minutes (except for the repeating messages). I can't even use CTRL-C or type at the root prompt.
I have stabilized the system by only specifying 50 VLANs.
We normally use VLAN numbers corresponding to internal project/customer numbers, so we use a random combination of very low VLAN IDs and very high VLAN IDs. I know it's possible to specify individual VIDs on the bridges as they change, but that's a huge pain.
---
Question #1:
My Boss is very mean, and demands to know why these Supermicro systems - which have been WORKING FINE WITH VMWARE for almost a decade - apparently have trouble doing basic VLAN routing with Proxmox, because of this problem. I know this isn't Proxmox's fault, but we've both been networking for a while, and never had to buy "beasty" network cards just to assign any VLAN IDs we wanted.
Question #2:
root@ci-proxmox04-cam-s:~# ethtool -k nic0 | grep vlan
rx-vlan-offload: on [fixed]
tx-vlan-offload: on [fixed]
This means I can't turn this feature off right?
Question #3:
Right now I am allowing only a range of 50 VLANs. How many can I allow before the memory becomes full? Is it trial and error?
Question #4:
Is there a danger of the card suddenly going crazy during production if I assign too many VLANs?
Question #5:
I do have access to the Enterprise Repo, but not on these particular older systems. Will the Enterprise Repo have updated drivers?
*** Thanks for any responses. This minor network glitch has really disrupted our rollout plans. ***
If you have a Proxmox subscription for the affected servers, you should create a support ticket, the support staff at proxmox is usually very helpful.
Apart from this, if your boss is indeed a "mean" guy, then maybe you should tell him not to waste his money (and your time) on old hardware. "Almost a decade" old hardware for production workload is ... frightening, to put it the least.