Skip to:

  1. Skip to Jira Navigation
  2. Skip to Side Navigation
  3. Skip to Main Content

ICMP availability with custom packet size doesn't work with JNI

Description

With the following configuration the services goes down with unknown reason. The service is configured in the poller-configuration.xml as follows:

Strange behavior can bee seen in the attached TCP dump files, the 1455 ICMP is not send from OpenNMS when it is configured to use the JniPinger.

JICMP has installed:

Acceptance / Success Criteria

None

Attachments

2

Lucidchart Diagrams

Activity

Benjamin Reed November 25, 2014 at 12:51 PM

Alright, I've fixed develop so it uses the latest 1.4.4 JICMP – if you have a good test environment please give it a shot with develop (once it's built) and let me know if you still see issues.

Benjamin Reed November 21, 2014 at 10:32 AM
Edited

You know, you changed the java code in jicmp, but we need to actually do a release of jicmp/jicmp6 including a 'maven deploy', and then change the dependency version in opennms-icmp-jni:

...and opennms-icmp-jni6:

So this really shouldn't be closed yet. Assigning to me to do release stuff.

fooker October 27, 2014 at 8:48 PM

fooker October 27, 2014 at 8:35 PM
Edited

The JniIcmpMessenger part receiving ICMP echo replies assums the default packet size of 56 bytes and drops everything else resulting in a time-out which is displayed as 'unknown'.

The check agains the packet size seems to be la left-over of ancient times and is covered by the JICMP packet parsers.

fooker October 27, 2014 at 7:42 PM

Fixed out of bounds error:
https://github.com/OpenNMS/jicmp/tree/jira/NMS-7073

BUT: there seems to be another problem, as the pollers are still reported as down.

Fixed

Details

Assignee

Reporter

Labels

Components

Fix versions

Affects versions

Priority

PagerDuty

Created October 22, 2014 at 5:02 AM
Updated May 11, 2015 at 2:49 PM
Resolved November 25, 2014 at 12:51 PM