snmp_profiles: fixes wrong values and unify OIDs

[FFHener]

• fix scale for bitrate of ports (example)
• collect out errors of interfaces as well (example with no errors)
• set correct type for interface errors and bitrate
• use the correct OID for uptime of tplink-switches (example of uptime working) and edgeswitches
• always start a OID with a .
• Always use the Description/Label of a Interface
  • this helps to know which Interface is what
  • this might resolve in some Interfaces not being reported correctly if they have the same names
  • for better readability of the monitoring data i prefer Descriptions

We might want to update the /etc/collectd/conf.d/bbb-snmp.conf file after this PR got merged for affected router to get correct data reported. This can be done without a sysupgrade by simply replacing this file and doing a reload of collectd.

[FFHener]

Im running again into Value too old errors so this will be a draft until i fixed it. There is a open Issue at collectd that would solve this. This happens when we have several instances of a name. One example is empty names for LAGs in Edgeswitches or not defined ports in tplink switches.

> snmpwalk -v 1 -c public fesev-switch.olsr .1.3.6.1.2.1.31.1.1.1.18
[...]
iso.3.6.1.2.1.31.1.1.1.18.10 = STRING: "dhcp"
iso.3.6.1.2.1.31.1.1.1.18.1000 = ""
iso.3.6.1.2.1.31.1.1.1.18.1001 = ""
[...]

> snmpwalk -v 1 -c public strom-poe.olsr .1.3.6.1.2.1.31.1.1.1.18
[...]
iso.3.6.1.2.1.31.1.1.1.18.1 = STRING: "core"
iso.3.6.1.2.1.31.1.1.1.18.2 = ""
iso.3.6.1.2.1.31.1.1.1.18.3 = ""
[...]

So we have the option of going back to index numbers or have some errors. I still need to check if this errors result in no data being sent in general. Maybe we can just tell collectd in general to stop reporting errors :D

[FFHener]

I decided for supressing all errors of collectd in favour of having nicer names in monitoring. This means label of ports that are used at a location should be unique

[noxilixon]

lgtm 👍