[prev in list] [next in list] [prev in thread] [next in thread] 

List:       opennms-buglist
Subject:    [Buglist] [Bug 3808] highThreshHold notifications do not work for data sources that do not have SNMP
From:       bugzilla () opennms ! org (bugzilla at opennms ! org)
Date:       2010-07-27 23:32:23
Message-ID: 20100727233223.A05D37884C7 () nen ! opennms ! org
[Download RAW message or body]

http://bugzilla.opennms.org/show_bug.cgi?id=3808


agalue at sync.com.ve changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




------- Comment #9 from agalue at sync.com.ve  2010-07-27 19:32 -------
Working on the solution for this Bug, I found a problem associated with the
solution proposed for the Bug 2711.

First we should know that the information about the Node/Interface/Service of
an event, in the case of thresholds executed by Collectd, are related to the
Collection Agent. So, the event's interface will contain the IP address of the
physical interface used for data collection.

This fact is so important that the Notification Manager uses this information
from the incoming event, to validate that it comes from a valid source; that's
the reason of why notifications are not working as expected right now.

The solution for the Bug 2711 overrides the event's interface content with the
IP address of the interface resource (only for thresholds that use interface as
resource type); which is wrong.

An interface resource could has (or not) an IP address associated with it; but
this address should not be confused with the IP address used on the event's
interface.

The current code introduces some changes to respect the initial fact. But, for
the people who are interested to know the IP address of the interface resource
affected by a threshold, the event will contain a new parameter called
ifIpAddress (only if the interface resource has an ipAddress associated with
it).

So, for interface resources, the threshold event will contain three additional
parameters:

ifIndex
ifLabel
ifIpAddress

The problem related to the bad rearm on interface resource thresholds, is
associated to the fact that the ds-label is usually not defined for this kind
of thresholds. If you check the reductionKey of the highThresholdExceeded
event, you will see that it needs the parameter "label" of the event. If this
is not configured, the default value is 'Unknown', and of course the auto-clear
feature won't work as expected.

I think that use the instance on reductionKey, could avoid this problem, but
now, if you don't define the ds-name for an interface threshold, the content of
"label" will be the same as "ifLabel".

This explanation is still valid for the problem described by the Bug 2711: "New
in-memory thresholder (collectd) does not handle non-IP interfaces".


-- 
Configure bugmail: http://bugzilla.opennms.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[prev in list] [next in list] [prev in thread] [next in thread]