Topic
  • 1 reply
  • Latest Post - ‏2012-05-08T07:06:11Z by SystemAdmin
SystemAdmin
SystemAdmin
430 Posts

Pinned topic first connection with ITNM

‏2012-05-03T12:02:10Z |
I prepared a proof of concept installation (all-in-one) of ITNM 3.9 on RedHat 5.8 under Vmware. Testing environment is a small LAN with few PC, one 3COM managed switch and Vmware server with some VMs.
From the begging i've got some problems, which couldn't understand.
1. SNMP Link state policy works unstable. It may detect link up/down state change, but not every time. I changed pooling period to 1 minute, but i am not sure this may be a cause.
itnm_poller.pl shows that this policy is enabled and it act on switch chassis.
2. In case link got down, RCA shows as root cause end node, not port in the switch. The same thing, when switch is powered off: root cause is indicated end node and the switch only symptom. After we configured switch to send SNMP traps and configured NCO ping probe, RCA became more "correct", but not at all. It looks like ITNM got confused with event generated by itself and "correct" works with snmp traps and ping probe alerts.
Pooling station is into discovery scope. Moreover, to be sure i populated NcpServerEntity into EventGatewaySchema.cfg with server ip address. Network hop view shows right L2 links between server, switch and end-nodes.
3. After i applied Fixpack 1, ITNM got problems with topology editor. When i try to create or delete any device or link manually in "Network Hop View", ncp_model precess frieze and only kill -9 on it helps.
Updated on 2012-05-08T07:06:11Z at 2012-05-08T07:06:11Z by SystemAdmin
  • SystemAdmin
    SystemAdmin
    430 Posts

    Re: first connection with ITNM

    ‏2012-05-08T07:06:11Z  
    I found by myself half of answer for the second question.
    I increased setup ncp_g_event process logging level to debug and i saw, that switch link down event was suppressed by endnode pingfail by the Connected Entity Suppression stitcher
    http://publib.boulder.ibm.com/infocenter/tivihelp/v8r1/topic/com.ibm.networkmanagerip.doc_3.9/itnm/ip/wip/event/concept/nmip_evnt_rcastitchersequence.html
    and event didn't arrive to the isolation stitcher
    I don't know why, but pingfail event arrives before linkdown event and it's fired as root cause.
    I saw into
    %NCHOME/precision/eventGateway/stitchers/RCA/ConnectedEntitySuppression.stch
    i found message
    This rule will NOT suppress an event on a connected interface entity if that interface is an Isolation Point.
    It looks like ITNM didn't identify switch link as a "Isolation Point". I didn't find where and how ITNM mark Entity as a isolation point.
    I topology map i see next L2 connections:
    itnm server - (int_1) Switch (int_24) - EndPoint
    From the diagram int_24 is isolation point, but why ITNM has another opinion?