Configuration & Administration

Expand all | Collapse all

Cambium 820 Custom Graphs

  • 1.  Cambium 820 Custom Graphs

    Posted 07-01-2018 04:03 AM
    I've built a customer monitoring template for a Cambium 820 radio system and when trying to create graphs, I only get N/A's....this is the ONLY set of graphs doing this.

    If I do a test on the OID's, I get the correct values but I cannot get them to graph, therefore I can't set triggers or anything along that line for the data either.

    I have noticed that if I SNMP walk the device and try and get all entries, it takes a very long time and eventually errors out.

    Any idea how to fix this? Or what log files would I need to dive into?

    ------------------------------
    Tim Meads
    NOC Supervisor
    Mountain West Technologies Network
    ------------------------------


  • 2.  RE: Cambium 820 Custom Graphs

    Posted 07-01-2018 04:29 AM
    I would suggest to take a look on snmp provider side first and ensure:
    • it's configured to serve snmp v2c
    • you can get values with snmpwalk command without issues

    Next you can try to tune zSnmpTimeout (10 sec by default) and ensure Zenoss is conigured to use snmp v2c - zSnmpVer property

    ------------------------------
    Serge
    ------------------------------



  • 3.  RE: Cambium 820 Custom Graphs

    Posted 07-02-2018 08:26 AM
    What error message when you walk the device?

    A few other zProps to note for issues polling devices using Snmp.

    zMaxOIDPerRequest - Try lowering if you are getting timeouts, default is 40.
    zSnmpTimeout
    zSnmpTries

    One thing to keep in mind, you have a default collection interval of 5 minutes (300 seconds).

    Adjusting these zProperties could make collection time even slower. Lowers the MaxOIDs makes Zenoss have to send more bulkGet requests. Timeout makes Zenoss wait longer to get a response. Tries will make zenoss try the bulkGet request again after a timeout.

    If you lower/raise these settings enough, you could force the collection time to take longer than the cycle time. If this happens, Zenoss will stop the collection and mark it as a missed run.

    You will see Info events like these:
    • Scan stopped; Collection time exceeded interval - Elapsed time 363.277658 seconds greater than 300 seconds
    • Missed run: 192.168.168.168


    ------------------------------
    jstanley
    ------------------------------



  • 4.  RE: Cambium 820 Custom Graphs

    Posted 07-03-2018 12:51 PM

    So the zMax would be lowered not highered?

    I can snmpwalk and test the template OID's from the template / device page, and I get the exact information I would actually get if it was functioning properly.

    I will try both of your suggestions.....the error that I'm referring too is that if I snmpwalk it from a command line on my computer, it spits out so much information and eventually errors out. 



    ------------------------------
    Tim Meads
    NOC Supervisor
    Mountain West Technologies Network
    ------------------------------



  • 5.  RE: Cambium 820 Custom Graphs

    Posted 07-03-2018 01:38 PM
    Again, with what error message?

    And yes, if you are getting unexplained timeouts. Sometimes devices can't handle large bulkGet requests.

    ------------------------------
    jstanley
    ------------------------------



  • 6.  RE: Cambium 820 Custom Graphs

    Posted 07-03-2018 02:22 PM

    So....*again*, Zenoss doesn't show me an errors......The graphs show N/A's......

    When I run the snmpwalk from my command line I get a timeout, which I ran against an word count to see how many lines are produced, and I ran it against a stop watch, Which went for 10 minutes 16 seconds before it timed out. 

    root@MacBookProTouch [~] : snmpwalk -v2c -cmWtCorp 10.58.0.89 .1 | wc

    Timeout: No Response from 10.58.0.89

       17642   72981 1225944

    If you're not familiar with WC, it's a command that counts words, lines, and bytes of a file or outputs of commands. So this shows me, 17,642 lines, 72,981 words in those lines totaling 1,225,944 bytes of text.......

    So I'll set my snmp time out to 600 seconds....and see what happens, but when I set it to 300, it didn't produce a zen error, nor did it gather the data in the graph....I would assume tho taking in this kind of data for 1 device (I currently have 8 I'm monitoring and will soon have 12 total, they are backhaul radios), that can add up to some serious crunching power......

    Oh and there's your timeout error...



    ------------------------------
    Tim Meads
    NOC Supervisor
    Mountain West Technologies Network
    ------------------------------



  • 7.  RE: Cambium 820 Custom Graphs

    Posted 07-03-2018 04:01 PM
    Did change a thing....I tried searching the logs for any error related to an IP of that unit and I can't find anything at all =\

    ------------------------------
    Tim Meads
    NOC Supervisor
    Mountain West Technologies Network
    ------------------------------



  • 8.  RE: Cambium 820 Custom Graphs

    Posted 07-03-2018 05:04 PM
    It seems only God, who knows about wc command, can help you.
    Thy God know everything about your setup, software versions, network environment and Cambium 820
    Suggest to pray for enlightenment or try to describe why your issue is related to zenoss (you have snmp timeouts)

    ------------------------------
    Serge
    ------------------------------



  • 9.  RE: Cambium 820 Custom Graphs

    Posted 07-04-2018 03:28 AM
    Hi Tim,
    You said originally that you had "tested OIDs" but I am getting the impression that a command-line snmpwalk is timing out?  Not a command run from the Zenoss GUI but a straight-forward unix command-line snmpwalk.?  If that doesn't work in a sensible length of time (and I don't think 10 minutes is "sensible"), then Zenoss doesn't have a chance.

    What are the OIDs in question?  What sort of data are you trying to get?  Are you actually trying to get data for "components" of the device, or data for the overall device itself?  If the former, then you are going to need to do some more work to build a modeler plugin in a ZenPack to model the components before you can collect data for them. (If this doesn't make much sense, then for a standard example, think of interfaces as a component of any snmp device).

    Another classic issue building a monitoring template is that if you are getting data for the overall device (eg. device up time), then in SNMP, this will be a scalar value - there's only one answer; if you are gathering data for a component, then snmp will have a table of values (for example, in standard MIB-2, interfaces is 1.3.6.1.2.1.2, ifTable is 1.3.6.1.2.1.2.2 and if Entry is 1.3.6.1.2.1.2.2.1, with values under that for ifIndex,  ifDescr, ifSpeed, etc).  If you are gathering a device scalar value, then in your template, you need the OID to end in 0; if you are gathering data from a component, using SNMP table data, then you need to arrange (in your modeler), that the index for the correct component is applied to get the appropriate value from the SNMP table.

    I don't think that this OID scalar / table stuff is causing your current problem - just that it might be the next issue to hit you.

    Perhaps you can post the OIDs you are trying to gather?

    The snmpwalk command you show is walking the entire MIB tree from your device.  I would do some more limited tests with a a command-line snmpwalk.  Check that you can get standard MIB-2 values (pretty well instantaneously):

    snmpwalk -v2c -cmWtCorp 10.58.0.89 .1 system

    If that works, start with some OIDs from your Cambium 820. From a quick Google, it looks like Cambium MIBs start 1.3.6.1.4.1.17713 (check I have the correct MIB here), so try:
    snmpwalk -v2c -cmWtCorp 10.58.0.89  1.3.6.1.4.1.17713

    I have seen some SNMP devices "get stuck" if they don't support part of a MIB so you may need to explore further down the Cambium OID tree.

    The Zenoss logfile for SNMP data collection is zenperfsnmp.log - but i doubt there will be anything more useful in there than a timeout report.

    Cheers,
    Jane

    PS. I am also a fan of wc, though I normally use the "-l" parameter to count lines.


    ------------------------------
    Jane Curry
    Skills 1st United Kingdom
    jane.curry@skills-1st.co.uk
    ------------------------------



  • 10.  RE: Cambium 820 Custom Graphs

    Posted 07-04-2018 09:34 AM
    Edited by Jay Stanley 07-04-2018 09:37 AM
    @Tim

    I originally asked for the walk error, not the Zenoss error. For us to best help you, it better to be more verbose with the errors you are getting.​

    Some information I would like to see.

    1. Post a screenshot of the datasource you are trying to graph.
    2. Post a screenshot of your graph's graph points
    3. Do your snmpwalk again, but this time without using wc, I would like to see what OID it times out on
      1. Test this more than once to see if it times out on the same OID or not
      2. Try walking the device from a few OIDs before the time out, this should tell you if there is an issue with the OID being requested or just the bulkGet of the snmpwalk
    4. Do a snmpwalk of the OID from the datasource you are trying to graph (Show command and output)
    5. execute 'zenperfsnmp run -v10 -d $DEVICEID |& tee /tmp/snmp_collection.out'
    6. Are you getting Snmp Agent Down events for the device? If so, did you try lowering the maxOIDs?

    I feel like this is something simple. Like a missing .0 at the end of the OID in the datasource.

    Please make sure to scrub any sensitive data from any screenshots or logs you post. And use pastebin.org or similar for any log data that is more than a few lines.

    ------------------------------
    jstanley
    ------------------------------


  • 11.  RE: Cambium 820 Custom Graphs

    Posted 07-07-2018 06:26 PM

    Jane,

    I appreciate the reply....

    Here is a screenshot of the OID's in the monitoring template I've created as well as a bit of the info....I had started to develop my graphs and such when I realized it wasn't pulling the info so I stopped. What I'm trying to pull is data from the radio in regards to signal strength, MSE, the modulation profile, and those kind of critical components.

    Cambium OIDsWhen I go into the device, and run a 'test' on the OID in zen, it comes back wtih the correct value right away.

    OID test

    I've tested this with all OID entries and they come back right away.

    On the other hand, running the test that you suggested of trying to gather just the enterprise OID"s, fails. Which is kinda shocking to be honest.

    root@MacBookProTouch [~] : snmpwalk -v2c -cmWtCorp 10.58.0.89  1.3.6.1.4.1.17713

    SNMPv2-SMI::enterprises.17713 = No Such Object available on this agent at this OID

    root@MacBookProTouch [~] :






    ------------------------------
    Tim Meads
    NOC Supervisor
    Mountain West Technologies Network
    ------------------------------



  • 12.  RE: Cambium 820 Custom Graphs

    Posted 07-08-2018 11:41 AM
    All your datasources start with 1.3.6.1.4.1.2281.10, but you did a snmpwalk using 1.3.6.1.4.1.17713? Is that a typo?

    When you do a snmp test in Zenoss, it shows you the correct OID.

    Your walk is against 1.3.6.1.4.1.2281.10.7.4.1.1.6, but nothing is returned for that OID, but instead the walks continues and returns data for an OID down the tree (1.3.6.1.4.1.2281.10.7.4.1.1.6.268451969)

    When you setup an snmp datasource in Zenoss, the OID you use works in two ways.

    First and always, Zenoss does a Snmp GET, not a WALK. The test datasource is very misleading, as it does a WALK.

    For device level templates, or templates on components without a snmpindex. The OID you provide, is the OID tries to pull for data. Which means, if you enter the OID 1.3.6.1.4.1.2281.10.7.4.1.1.6.2 it will try to pull that, but not 1.3.6.1.4.1.2281.10.7.4.1.1.6.268451969.

    But if a template lives on a component and the component has a value set to snmpindex, then the behavior is changed slightly.

    Instead of doing a get for 1.3.6.1.4.1.2281.10.7.4.1.1.6, Zenoss will instead append the snmpindex to the end of the OID set in the datasource. So, if a component had a snmpindex of 2, the OID Zenoss would try to get would be 1.3.6.1.4.1.2281.10.7.4.1.1.6.2.

    This is done so that you can use a base OID, but get different values for each component. Think Interface data.


    ------------------------------
    jstanley
    ------------------------------



  • 13.  RE: Cambium 820 Custom Graphs

    Posted 07-09-2018 04:31 AM
    Looking at a description of the MIB here -  http://www.circitor.fr/Mibs/Html/M/MWRM-RADIO-MIB.php the data that you are trying to get is tabular (ie down a table -> entry construct in the SNMP OID) so, translating that to the Zenoss  environment, you need to create a ZenPack that will model the appropriate component (radio????) and arrange that the snmpindex attribute is the correct index into your snmp table.

    If you've not done anything like this before, you could have a look at the ZenPackdevelopers' Guide - https://github.com/ZenossDevGuide/DevGuide   - chapter 9 is about creating an SNMP ZenPack including a modeler.

    As Jay says, although template tests do an snmwalk, the actual code that gets data values for storage and graphing does an snmpget - and it needs the correct index on the end.

    Cheers,
    Jane

    ------------------------------
    Jane Curry
    Skills 1st United Kingdom
    jane.curry@skills-1st.co.uk
    ------------------------------