is_tag in table does not suppress field value

[candlerb]

Relevant telegraf.conf:

Stock default

System info:

telegraf 1.6.3-1, Ubuntu 16.04

Steps to reproduce:

[[inputs.snmp]]
  agents = [ "x.x.x.x" ]
  version = 2
  community = "public"
  interval = "60s"
  timeout = "10s"
  retries = 3

  [[inputs.snmp.field]]
    name = "hostname"
    oid = "RFC1213-MIB::sysName.0"
    is_tag = true

  [[inputs.snmp.field]]
    name = "uptime"
    oid = "DISMAN-EXPRESSION-MIB::sysUpTimeInstance"

  # IF-MIB::ifTable contains counters on input and output traffic as well as errors and discards.
  [[inputs.snmp.table]]
    name = "interface"
    inherit_tags = [ "hostname" ]
    oid = "IF-MIB::ifTable"

    # Interface tag - used to identify interface in metrics database
    [[inputs.snmp.table.field]]
      name = "ifDescr"
      oid = "IF-MIB::ifDescr"
      is_tag = true

Then

telegraf --test --config /etc/telegraf/telegraf.d/sw1.conf

Expected behavior:

ifDescr would appear as a tag, but not as a value (since that would be unnecessary duplication of data)

Actual behavior:

...
> interface,agent_host=x.x.x.x,host=tick,ifDescr=24\ Gigabit\ -\ Level,ifIndex=24 ifInNUcastPkts=22883363i,ifOutOctets=1203804267i,ifOutUcastPkts=271421313i,ifOutErrors=0i,ifMtu=1518i,ifSpeed=1000000000i,ifOperStatus=1i,ifInUcastPkts=280321307i,ifOutNUcastPkts=26088154i,ifOutDiscards=87i,ifType=6i,ifAdminStatus=1i,ifInErrors=0i,ifInUnknownProtos=0i,ifInOctets=1187323181i,ifDescr="24 Gigabit - Level",ifPhysAddress="a0:04:60:26:4f:f8",ifLastChange=1306698600i,ifInDiscards=0i 1527344674000000000

ifDescr appears as both a tag and a value.

Having both tag and value can be dealt with, e.g. in Chronograf you need to write queries like WHERE ifDescr::tag=:ifDescr: (where :ifDescr: is a template variable) - it's the ::tag which caught me out. But I think it would be better to suppress the non-tag field entirely.

Additional info:

N/A

[danielnelson]

As a workaround you should be able to use fielddrop to remove the field variant.

@phemmer This feels like a bug to me, do you know of any reason we shouldn't adjust this behavior?

[phemmer]

Yes, this is a bug. From a brief glance in the code, it looks like how this happens is that we first take the preconfigured [[inputs.snmp.table.field]] and add it to the list of OIDs to fetch. We then automatically configure all the fields on the [[inputs.snmp.table]], and the OID gets added again.