VNM.OUT is huge due to tons of read_attr error

Products

• Spectrum
• Network Observability

Issue/Introduction

In the SpectroSERVER $SPECROOT/SS/VNM.OUT file there are tons of these entries:

read_attr error: No value on the scratchpad returning Internal Failure !!

The problem is occurring on the primary and secondary SSs in a DSS environment.

 

This is an example of the VNM.OUT file size:

$ ls -l --block-size=MB | grep VNM

-rw-r--r-- 1 spectrum spectrum 3014MB Jan 22 03:18 VNM.OUT

 

This is an example of the number of error entries in the VNM.OUT file:

$ cat VNM.OUT | grep 'read_attr error: No value on the scratchpad returning Internal Failure !!' | wc -l

13523847

Environment

Spectrum 22.2.x and 23.3.x

Resolution

Add the following row/line in the $SPECROOT/SS/.vnmrc file.

enable_instancelist_watch_scratchpad_fix=true

Then reload the Spectrum VNM .vnmrc resource file using CLI action=0x10700 logged as Spectrum Install Owner account.

$ cd $SPECROOT/vnmsh

$ ./connect

$ ./update action=0x10700 mh=<VNM model_handle>

$ ./update action=0x10700 mh=0x1000000     (example - for landscape 0x1000000)

$ ./disconnect

Additional Information

https://techdocs.broadcom.com/us/en/ca-enterprise-software/it-operations-management/spectrum/23-3/release-information/issues-resolved.html

Symptom:  When a user uses the Spectrum watch functionality with multiple levels of watches and monitors all the instances of the interface, the following error is logged in VNM.out:
"read_ttr error: No value on the scratchpad!!"
 
Resolution:  With this fix, Spectrum provides a new attribute enable_watch_scratchpad_error_report  in the .vnmrc file to address the issue of unnecessary logs. This issue typically happens because there are some instances for which a few attributes have no data on the scratchpad. As a result, the "read_attr error: No value on the scratchpad!!" error is logged in VNM.out for such instances. The fix now resolves this issue. 
(DE569058, 33460557, 23.3.1)