False Fault Isolation Unresolved Fault Detected Alarms

Products

• CA Spectrum
• CA eHealth

Issue/Introduction

We are still seeing Unresolved Fault Detected Alarms against Fault Isolation models that cannot be cleared.  We have set the Fault Isolation setting to Device in Fault Domain but these alarms generate on the FaultIsolation model:

 

Critical May 6, 2020 1:01:28 PM CDT Fault Isolation "Fault Isolation has detected an unresolved fault (fault record id 3305).  The following is a list of models to which SPECTRUM has lost contact:

192.168.10.80 with a device criticality of 0
192.168.10.90 with a device criticality of 0
192.168.10.920 with a device criticality of 0

As a result of this condition, an alarm has been generated on this model." System May 6, 2020 1:01:28 PM CDT System 0x10d05 10 FaultIsolation

Environment

Release : 10.4

Component : Spectrum Core / SpectroSERVER

Cause

WA_Link and WA_Segment modeling is not correct.  The following possibilities can cause this:

1.  There were WA_Segment models connected to ports but the WA_Segment model did not have a corresponding WA_Link.  For proper modeling there should be a WA_Link model that contains the WA_Segment model and allows for the fault isolation algorithm to run.

2.  The WA_Segments are connected to device models. The WA_Segment is intended to be connected to interface models only.

Resolution

Run a Locater search for ModelType Name Equal to WA_Link and run another for ModelType Name Equal To Wa_Segment. Compare the results.  There should be 1 WA_Link per WA_Segment and vice versa.  We found here on a couple of the SS, there were 0 WA_Links and 200 WA_Segments.

To resolve this, destroy all WA_Segments that do not have an associated WA_Link.

Likewise, if the WA_Segment is connected to the device, destroy both the WA_Segment and the associated WA_Link.