Odd syslog kernel message: selects TX queue 2, but real number of TX queues is 2

[Cupidazul]

whenever I start gwlbtun I am seeing syslog consecutive messages, flooding my log file with:

May  6 10:22:40 ip-10-56-132-4 kernel: [ 1952.775858] ens5 selects TX queue 2, but real number of TX queues is 2
May  6 10:22:40 ip-10-56-132-4 kernel: [ 1952.776814] ens5 selects TX queue 2, but real number of TX queues is 2
May  6 10:22:41 ip-10-56-132-4 kernel: [ 1953.777157] ens5 selects TX queue 2, but real number of TX queues is 2
May  6 10:22:41 ip-10-56-132-4 kernel: [ 1953.778146] ens5 selects TX queue 2, but real number of TX queues is 2
May  6 10:22:42 ip-10-56-132-4 kernel: [ 1954.778445] ens5 selects TX queue 2, but real number of TX queues is 2
May  6 10:22:42 ip-10-56-132-4 kernel: [ 1954.779273] ens5 selects TX queue 2, but real number of TX queues is 2
May  6 10:22:43 ip-10-56-132-4 kernel: [ 1955.779642] ens5 selects TX queue 2, but real number of TX queues is 2
May  6 10:22:43 ip-10-56-132-4 kernel: [ 1955.780513] ens5 selects TX queue 2, but real number of TX queues is 2
.
.
.

do you have any pointers on why this happens or if its not important, can I disable this from flooding syslog?

my usecase, using:
create-passthrough.sh
with "#define NO_RETURN_TRAFFIC" enabled
Ubuntu 20.04.6 LTS
vmlinuz-5.15.0-1058-aws
t3.micro

[grayaw]

What version of the ENA driver are you using with that image? Older ENA drivers had issues with queue selection with packets received from virtual TUN/TAP interfaces (which gwlbtun uses), fixed in ENA 2.9.0 or later.

You can see the version loaded in either dmesg, or by doing "ethtool -i (network interface)". You will see something like:

$ ethtool -i ens5
driver: ena
version: 2.11.0g
(additional stuff)

[Cupidazul]

ethtool -i ens5
driver: ena
version: 5.15.0-1058-aws
firmware-version:
expansion-rom-version:
bus-info: 0000:00:05.0
supports-statistics: yes
supports-test: no
supports-eeprom-access: no
supports-register-dump: no
supports-priv-flags: no

[grayaw]

Hrm. That's just the kernel version, interesting. Any chance you can grab the current ENA driver from https://github.com/amzn/amzn-drivers/tree/master/kernel/linux/ena and see if that resolves the errors?