Open welkin-he opened 1 month ago
CSCwf58763 specifically says to flag for DNs for node IDs that exist, but may be currently in a different pod compared to its current location.
"In this example, Node 401/402 are Spines in Pod 2 but they are stale as if they are in Pod 1.
node-1# moquery fabricRsDecommissionNode | egrep "^dn|^_svc"
_svc : policydist
dn : uni/fabric/outofsvc/rsdecommissionNode-[topology/pod-1/node-402]
_svc : policydist
dn : uni/fabric/outofsvc/rsdecommissionNode-[topology/pod-1/node-401]
_svc : policymgr
dn : uni/fabric/outofsvc/rsdecommissionNode-[topology/pod-1/node-401]
_svc : policymgr
dn : uni/fabric/outofsvc/rsdecommissionNode-[topology/pod-1/node-402]
"
cool, thanks for putting two validation together. Here is the initial commit for peer view. https://github.com/datacenter/ACI-Pre-Upgrade-Validation-Script/tree/welkin-issue-164
(use upvote :thumbsup: for attentions)
Validation Type
[x] - Bug
What needs to be validated
fabricRsDecommissionNode mo with node-id match any existing spine switches, while upgrading to 5.2(3e) and above unitl 6.0(3d)
Why it needs to be validated
The stale mo can trigger spine get removed from coop adjacency list for all leaf switches, while keep the spine participating forwarding decision without the latest insight of the fabric.
Additional context
Add any other context about the feature request here.