NewValidation: Check for stale APIC entries

[pcelarek]

(use upvote 👍 for attentions)

Validation Type

[ ] - Other

What needs to be validated

We have recently faced a problem with a stale APIC entry that was there after APIC cluster size change that was done few years ago. After the APIC upgrade this entry caused some issues within the fabric and created false positive F1543.

There should be a check as part of pre-validation that would look for this condition so it can be fixed by TAC before proceeding with APIC upgrade.

We should validate for APICs in "ERASED" state that indicates stale entry.

Example:

APIC: avread:
Cluster:
[...]

APICs:
[...]

                  APIC 5                  
version                                  
address           X.X.X.X            
oobAddress        Y.Y.Y/Y        
routableAddress   0.0.0.0                 
tepAddress        0.0.0.0                 
podId             0                       
chassisId         035ff932-.-250d8273     
**cntrlSbst_serial  (ERASED,)**               
active            NO (zeroTime)           
flags             ----              
health           0

It could be also part of already existing check to avoid creating new check:
[Check 6/64] APIC Cluster is Fully-Fit... PASS

[monrog2]

Hello!

It is expected to see ERASED controllers via avread, so having a check based solely on that may lead to many false positive:
https://bst.cloudapps.cisco.com/bugsearch/bug/CSCvn12926

You mention that this led to other issues down the line. If you are able to share additional details on those issues, or send me an email directly with those details/sr number ([email protected]), I can dig to see if there is something more specific that can be flagged here related to the symptoms it may have caused.