When creating SpectroWatches with multiple levels of watches for monitoring all of the instances of an interface after a few poll cycles we see the following error:
The evaluation of watch <watch_name> failed for model <model_name>: SPC-WCH-20072:This is a user defined error.
The recovery and failure of the evaluation repeats periodically every two/three polling intervals.
Release : 10.4.2.1 / 10.4.2.2
Component : Spectrum Core / SpectroSERVER Watches
The scratchpad data becomes null and the watches will fail.
For Spectrum 10.4.2.1 please request patch d74
For Spectrum 10.4.2.2 please request patch d88
For Spectrum 10.5 / 21.2.1 there is a VNMRC setting that can be added
enable_instancelist_watch_scratchpad_fix=true
This has been included out of box in Spectrum 21.2.1 and only .vnmrc setting is required.
Symptom: When I create a filter with multi-level watches to monitor all the instances of any attribute, in very rare-case due to missing data in the scratchpad, DX NetOps Spectrum watches exit abruptly with a user-defined error.
Resolution: Code changes are made to set enable_instancelist_watch_scratchpad_fix to true to re-load the scratchpad with the latest data so that the watch evaluation executes properly. (DE485740, 21.2.1)