[erlang-bugs] SNMP erlNodeReductions defined too narrow

[Tobias Schlager]

Hi all,

I just came across this line in the logs of a testing server running R16B01:
User error: Got 508191824798 from {otp_mib,erl_node_table,[]}. Using wrongValue

A SNMP walk with snmpwalk -v 2c -c public host:port .iso.org.dod.internet.private.enterprises fails with the following output:

...
SNMPv2-SMI::enterprises.193.19.3.1.2.1.1.1.3.1 = STRING: "BEAM"
SNMPv2-SMI::enterprises.193.19.3.1.2.1.1.1.4.1 = STRING: "5.10.2"
SNMPv2-SMI::enterprises.193.19.3.1.2.1.1.1.5.1 = Gauge32: 0
SNMPv2-SMI::enterprises.193.19.3.1.2.1.1.1.6.1 = Counter32: 99993630
SNMPv2-SMI::enterprises.193.19.3.1.2.1.1.1.7.1 = Counter32: 419219062
Error in packet.
Reason: (genError) A general failure occured
Failed object: SNMPv2-SMI::enterprises.193.19.3.1.2.1.1.1.7.1

Examining the exact value on the machine with erlang:statistics(wall_clock) confirms the wallclock value of 419219062 and I can't see a problem with that value.

However, when I look at the next value 193.19.3.1.2.1.1.1.8.1 (erlNodeReductions) manually, erlang:statistics(reductions) gives me a total number of 508213447798 reductions. This looks exactly like the value in the error message. Furthermore, the value is clearly out of range for a 'Counter32' which is the defined type for erlNodeReductions. Googling the topic brought up the following 3 year old repository https://github.com/dreverri/erlang_snmp_reduction_issue that contains code the reproduce this behaviour.

Regards
Tobias