SpectroSERVER runs out of memory and crashes when modeling XpedSwRtr models that support LLDP

Document ID : KB000006391
Last Modified Date : 14/02/2018
Show Technical Document Details
Issue:

When modeling XpedSwRtr models that support LLDP, the SS may run out of memory and crash.  The VNM.OUT will show a stack that has the following lines:

 

libadcor.so.1:__1cPIHLldpNeighborsQreadLldpRemTable6MrknNCsModelHandle_3rnNCsULHashTable_rknODeviceLLDPInfo__i_+494

libadcor.so.1:__1cPIHLldpNeighborsKgatherInfo6MrknNCsModelHandle_3_pnNCsAttrValList__+100

libadcor.so.1:__1cPIHLldpNeighborsQaction_triggered6MrknNCsModelHandle_3pknICsAction__pnNCsAttrValList__+9c

Cause:

The problem is due to having time filters enabled in the LLDP tables on the device.  To determine which device is the culprit, you can add the following entry to the <SPECROOT>/SS/.vnmrc file and stop/restart the SS for the change to take effect:

debug_enterasys_timefilter=TRUE

Run the discovery and take note of the model handle of the last entry in the output.

Resolution:

A temporary workaround may be to disable the LLDP time filters on the device:

set snmp timefilter break


If this is not possible or does not resolve the issue due to the number of entries in the LLDP table, then CA engineering has created a fix for the XpedSwRtr modeltype.  The code will skip the extensive timefilter table reads in the LLDP tables. This fix is available in debug patch  Spectrum_10.02.00.D156.  This fix is also tentatively scheduled to be included out of the box with the next release of CA Spectrum after 10.2.1.

If you need Spectrum_10.02.00.D156 please open a case with CA Spectrum Support.  The release notes show:

The SS starts consuming memory and runs out during discovery of LLDP. 

(DE274502, 00631083)