SRIOV Network Metric 'sriov_vf_tx_errors' is missing

[dipti262]

Hi,

We are using the NIC driver - i40e. We are not seeing the metrics - sriov_vf_tx_errors on the same node. Even though the other metrics are coming fine,

sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:00.2",pf="enp175s0f0",vf="0"} 37934
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:00.3",pf="enp175s0f0",vf="1"} 37934
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:00.4",pf="enp175s0f0",vf="2"} 37934
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:00.5",pf="enp175s0f0",vf="3"} 37934
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:00.6",pf="enp175s0f0",vf="4"} 37934
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:00.7",pf="enp175s0f0",vf="5"} 37934
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:01.0",pf="enp175s0f0",vf="6"} 37934
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:01.1",pf="enp175s0f0",vf="7"} 37934
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:02.2",pf="enp175s0f1",vf="0"} 44610
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:02.3",pf="enp175s0f1",vf="1"} 44610
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:02.4",pf="enp175s0f1",vf="2"} 44610
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:02.5",pf="enp175s0f1",vf="3"} 44610
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:02.6",pf="enp175s0f1",vf="4"} 44610
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:02.7",pf="enp175s0f1",vf="5"} 44610
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:03.0",pf="enp175s0f1",vf="6"} 44610
sriov_vf_rx_bytes{numa_node="1",pciAddr="0000:af:03.1",pf="enp175s0f1",vf="7"} 44610
# HELP sriov_vf_rx_dropped Statistic rx_dropped.
# TYPE sriov_vf_rx_dropped counter
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:00.2",pf="enp175s0f0",vf="0"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:00.3",pf="enp175s0f0",vf="1"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:00.4",pf="enp175s0f0",vf="2"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:00.5",pf="enp175s0f0",vf="3"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:00.6",pf="enp175s0f0",vf="4"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:00.7",pf="enp175s0f0",vf="5"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:01.0",pf="enp175s0f0",vf="6"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:01.1",pf="enp175s0f0",vf="7"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:02.2",pf="enp175s0f1",vf="0"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:02.3",pf="enp175s0f1",vf="1"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:02.4",pf="enp175s0f1",vf="2"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:02.5",pf="enp175s0f1",vf="3"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:02.6",pf="enp175s0f1",vf="4"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:02.7",pf="enp175s0f1",vf="5"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:03.0",pf="enp175s0f1",vf="6"} 0
sriov_vf_rx_dropped{numa_node="1",pciAddr="0000:af:03.1",pf="enp175s0f1",vf="7"} 0
In the pod log I am seeing this:

2025/06/13 08:05:15 enp175s0f1 - using netlink collector
2025/06/13 08:05:15 0000:dc:00.0 - could not get pf interface name in path '0000:dc:00.0/net'
open 0000:dc:00.0/net: no such file or directory
2025/06/13 08:05:15 error getting vf address
no virtual functions found for pf '0000:dc:00.0'
2025/06/13 08:05:15  does not support netlink collector
2025/06/13 08:05:15  - using sysfs collector
The collector order is --collector.vfstatspriority=sysfs,netlink

The same metrics sriov_vf_tx_errors is coming fine on another node having same i40e NIC driver.
I am not sure whether the NIC driver type is related to the metrics getting missed.

eccd@cp-bm08e2:~>  readlink -f /sys/class/net/enp175s0f0/device/driver | awk -F'/' '{print $NF}'
i40e
# TYPE sriov_vf_tx_errors counter
sriov_vf_tx_errors{numa_node="1",pciAddr="0000:d8:02.0",pf="ens3f0",vf="0"} 0
sriov_vf_tx_errors{numa_node="1",pciAddr="0000:d8:02.1",pf="ens3f0",vf="1"} 0
sriov_vf_tx_errors{numa_node="1",pciAddr="0000:d8:02.2",pf="ens3f0",vf="2"} 0
sriov_vf_tx_errors{numa_node="1",pciAddr="0000:d8:02.3",pf="ens3f0",vf="3"} 0
sriov_vf_tx_errors{numa_node="1",pciAddr="0000:d8:02.4",pf="ens3f0",vf="4"} 0

[bersoare]

do you see the dropped/errors counters as non zero when using ip -s l sh dev <pfname>?

im looking at a mellanox connectx 4, and i don't see a vf tx error counter available. it could be driver specific - the driver probably dont spit out the counters youre looking for:

 enp2s0f0np0: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc mq master bond0 state UP mode DEFAULT group default qlen 1000
    link/ether 0c:42:a1:8f:97:02 brd ff:ff:ff:ff:ff:ff
    RX:  bytes packets errors dropped  missed   mcast
    1971761474 4986347      0       0       0 1947390
    TX:  bytes packets errors dropped carrier collsns
     393429377 1737359      0       0       0       0
    vf 0     link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff, spoof checking off, link-state auto, trust off, query_rss off
    RX: bytes  packets  mcast   bcast   dropped
      55253301   464403  464125       0        0
    TX: bytes  packets   dropped
          5976       84        0
/sys/class/net/enp2s0f0np0/device/sriov/0# cat stats
tx_packets    : 84
tx_bytes      : 5976
tx_dropped    : 0
rx_packets    : 464403
rx_bytes      : 55253301
rx_broadcast  : 0
rx_multicast  : 464125
tx_broadcast  : 0
tx_multicast  : 84
rx_dropped    : 0

02:00.0 Ethernet controller: Mellanox Technologies MT27710 Family [ConnectX-4 Lx]