netlink speed check fail , "invalid argument"

when netlinks option is configured , nagios calls check_netlink in remote host to check interface information

# nagios command
/usr/lib/nagios/plugins/check_nrpe -H $HOSTADDRESS$ -c check_neutron_public -t 10

currently check_netlinks.py check link mtu and speed by reading files from /sys/class/net/<interface>/ directory, this works for most interfaces but not for some virtual-interface that has no speed information

For example , in Kubernetes deployment that use Juniper contrail CNI , contrail charm will create a vhost0 interface which has mtu but no speed information

19: vhost0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9000 qdisc fq_codel state UNKNOWN mode DEFAULT group default qlen 1000
    link/ether 48:df:37:9b:90:68 brd ff:ff:ff:ff:ff:ff

ubuntu@iadaz01skwrk01-k8s:~$ cat /sys/class/net/vhost0/mtu
9000

ubuntu@iadaz01skwrk01-k8s:~$ cat /sys/class/net/vhost0/speed
cat: /sys/class/net/vhost0/speed: Invalid argument

In this case , the check_netlinks.py will fail

ubuntu@iadaz01skwrk01-k8s:~$ python /usr/local/lib/nagios/plugins/check_netlinks.py --iface vhost0 --operstate unknown --mtu 9000
CRITICAL: vhost0 (speed returns invalid argument)

Although not function impacting, Nagios will show critical alarm on this interface

One potential solution is to do speed check only when the option is specified in nrpe netlinks config options, or give user option exclude certain link from speed check

Relevant nrpe config:

ubuntu@fce-tcspreprod:~$ juju config nrpe netlinks
- vhost0 mtu:9000 op:unknown
- < other interface config, skipped >