snmppysnmp

Use Pysnmp to update SNMP Table dynamically from agent side

I had been searching a way to use pysnmp to dynamically update a SNMP table during SNMP Agent running. But so far no luck...

The table had been already defined in MIB file(see below), but it seems I need to overwrite its "readGet()" method in order to return correct data from current system status.

According to the instructions I'm able to build a static table with pre-defined value before SNMP Agent startup, and after SNMP Agent startup:

# Register an imaginary never-ending job to keep I/O dispatcher running forever
self.snmpEngine.transportDispatcher.jobStarted(1)
# Run I/O dispatcher which would receive queries and send responses
try:
    self.snmpEngine.transportDispatcher.runDispatcher()
except:
    self.snmpEngine.transportDispatcher.closeDispatcher()
    raise

it is able to return my expected value.

But for my system, it will dynamically generate many alarm information, and those information need to be updated into SNMP's MIB Table, which allow other SNMP Manager to send "get/getNext" to fetch alarm information from my system.

So I would like to know

if so, how about getNext query? does it will ignore those empty table row? or it always return next one, even it is a dummy one?

@Ilya Etingof, the expert of pysnmp. Could you help me with it if you have time?

Br, -Dapeng Jiao

1.

MIB file definition of that alarm table (some sensitive info are removed)

alarmTable = MibTable((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3))
alarmEntry = MibTableRow((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1)).setIndexNames((0, "MY-MIB", "alarmIndex"))
alarmIndex = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 1), Integer32().subtype(subtypeSpec=ValueRangeConstraint(1, 2147483647))).setMaxAccess("readonly")
alarmId = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 2), Integer32().subtype(subtypeSpec=ValueRangeConstraint(1, 2147483647))).setMaxAccess("readonly")
alarmName = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 3), DisplayString().subtype(subtypeSpec=ValueSizeConstraint(0, 255))).setMaxAccess("readonly")
alarmSeverity = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 4), AlarmSeverity()).setMaxAccess("readonly")
alarmTime = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 5), DateAndTime()).setMaxAccess("readonly")
alarmType = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 6), AlarmType()).setMaxAccess("readonly")
alarmSource = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 7), DisplayString().subtype(subtypeSpec=ValueSizeConstraint(0, 255))).setMaxAccess( "readonly")
alarmCategory = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 8), DisplayString().subtype(subtypeSpec=ValueSizeConstraint(0, 255))).setMaxAccess("readonly")
alarmProbableCause = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 9), ProbableCause()).setMaxAccess("readonly")
alarmComparable = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 10), DisplayString().subtype(subtypeSpec=ValueSizeConstraint(0, 255))).setMaxAccess("readonly")
alarmAdditionalText = MibTableColumn((1, 3, 6, *, *, *, *, *, *, *, *, *, , 3, 1, 11), DisplayString().subtype(subtypeSpec=ValueSizeConstraint(0, 255))).setMaxAccess("readonly")
Solution

  • It is possible to maintain dynamic SNMP table with pysnmp. There are many approaches to this problem:

    1. Periodically update your table (via a table callback or dedicated thread) by calling mibInstrumentation.writeVars in this example script. The backsides includes delays in serving new data and queries if they come at the moment of running update. But no much coding required.

    2. Extend MibTableColumn class and implement its readGet/readGetNext methods to look into your data when being called and return OID/value pair. The complication here is that to handle GETNEXT query you need to maintain some sort of consistent ordering of OIDs and search for next greater than given.

    3. Ditch the whole pysnmp's SMI infrastructure and implement your own MIB controller on top of whatever data source you have to read data from. You would need to implement readGet (easy) and likely readGetNext (more complicated as stable OIDs sort is required) methods. This way you would relief yourself from learning the details of rather generic and complicated pysnmp SMI implementation and focus on your minimal requirements.

    To answer your other questions: