This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.
You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

Orion NPM response times not matching up with manual command prompt testing

FormerMember
FormerMember

Seeing something strange with my NPM that I'm wondering if anyone can shed some light on.

Basically, we have a location that almost all of the nodes are reporting above average response time...say 300ms. When I open a command prompt and ping the same devices, I see normal response times of around 75ms. I can let the command prompt pings go for hours and never see anything resembling these 300ms response times reported in NPM.

How does NPM perform response time calculations? Should I be able to see the same/similar times when manually pinging from a command prompt? Has anyone seen anything like this before?

Thanks for your help!

  • Are you opening a command prompt on the Polling Engine that the node is monitored by, or from your desktop?

  • FormerMember
    0 FormerMember in reply to jtimes

    Manual testing is being performed from the actual polling engine server, which is the same device that Orion is installed on. My thought is that if the slow response is valid, I should also see it during this test - which isn't happening.

  • okay, I don't have an answer for what you are seeing, only a few more suggestions to check.  Have you checked the Network Tab on the NPM to see if your "over stuffing" the ping packets.  Do a tracert, are the ping taking the "expected" path,  check the path that Orion is taking to ping those 300ms devices, check for utilization check the QOS buckets if implemented.  

  • FormerMember
    0 FormerMember in reply to jtimes

    Where are you suggesting I go for the Network Tab in NPM? If I go to File > Orion Network Performance Monitor Settings > Network tab, I only see limited network settings. Is there somewhere else I should look?

    Traceroutes show the expected path, but is there a way to see if NPM is taking a different route? I'm not suggesting this is possible, but if it is, it could explain the discrepancy from NPM to my manual testing.

    We are not using QOS buckets, but if we were, where is this configured within NPM?

  • FormerMember
    0 FormerMember

    Seeing something strange with my NPM that I'm wondering if anyone can shed some light on.

    Basically, we have a location that almost all of the nodes are reporting above average response time...say 300ms. When I open a command prompt and ping the same devices, I see normal response times of around 75ms. I can let the command prompt pings go for hours and never see anything resembling these 300ms response times reported in NPM



     

    We are seeing a very similar situation.  We have 20+ devices (routers, switches, servers) located in the same facility that started showing 300+ms response time several weeks ago.  The response time varies for all these devices between 190-300+ms.  When you (continuous) ping them manually from the Orion server the response times are 6ms.  Trace routes reveal the proper path is being taken. The data portion for ICMP packets is set to 0 bytes in Orion.  No one is complaining of slowness but I'm pulling my hair out on this one.  We're still running 8.5.1,

  • FormerMember
    0 FormerMember in reply to FormerMember

    Where do you confirm/set the data portion for ICMP packets (the 0 bytes you refer to)?

    Pretty much the same here...no one is complaining that things are slow, but management sees the slow response time in the reports and wants to know what is up.

  • FormerMember
    0 FormerMember in reply to FormerMember

    Where do you confirm/set the data portion for ICMP packets (the 0 bytes you refer to)?



    Try System Manager>Settings>Network.

  • FormerMember
    0 FormerMember in reply to FormerMember

    That's what I thought. I checked there and had some text in the field, but now that I removed it the problem remains.

    Any other ideas? We continue to see vast differences between command line ping/response times and those that Orion NPM is doing on the backend.

    Thanks for your help!

  • When is the last time you did the Polls Per Second Tuning on your NPM?


    What does you NPM Polling status look like?  on the NPM File>Polling Status


    Here is one of mine:


    Orion Network Performance Monitor Version 9.0.0
    NetPerfMon Engine 
    Network Node Elements 1028
    Interface Elements 4317
    Volume Elements 27
    Date Time 10/21/2008 9:57:44 AM
    Paused False
    Max Outstanding Polls 90
    Status Pollers 
    ICMP Status Polling Index 5372 out of 5372
    SNMP Status Polling Index 5372 out of 5372
    ICMP Polls per second 0
    SNMP Polls per second 64
    Max Status Polls Per Second 76
    Packet Queues 
    DNS Outstanding 7
    ICMP Outstanding 0
    SNMP Outstanding 90
    Statistics Pollers 
    ICMP Statistic Polling Index 9789 out of 9789
    SNMP Statistic Polling Index 2349 out of 9789
    ICMP Polls per second 0
    SNMP Polls per second 69.5
    Max Statistic Polls Per Second 76


  • FormerMember
    0 FormerMember in reply to jtimes

    It's probably been a few months since I did tuning. I'm not opposed to doing it again, but the response time issues I'm seeing are only affecting one site (about 20 devices).

    Orion Network Performance Monitor Version 8.5.1
    NetPerfMon Engine    
    Network Node Elements    379
    Interface Elements    8615
    Volume Elements    267
    Date Time    10/21/2008 10:21:08 AM
    Paused    False
    Max Outstanding Polls    900
    Status Pollers    
    ICMP Status Polling Index    9261 out of 9261
    SNMP Status Polling Index    1895 out of 9261
    ICMP Polls per second    0
    SNMP Polls per second    191
    Max Status Polls Per Second    191
    Packet Queues    
    DNS Outstanding    0
    ICMP Outstanding    0
    SNMP Outstanding    721
    Statistics Pollers    
    ICMP Statistic Polling Index    17174 out of 17174
    SNMP Statistic Polling Index    8446 out of 17174
    ICMP Polls per second    0
    SNMP Polls per second    130
    Max Statistic Polls Per Second    130