Upgrade from 3.2.4 to 4.2.3. resulting in a number of T1 showing DHCP overlapping errors. T1 showing as 'Failed' state - Error Code 22008

Products

• VMware NSX

Issue/Introduction

• Post upgrade from 3.2.4 to 4.2.3, the per-gateway DHCP server realization error seen because ip-pool overlapping reported at backend.
• This issue happens in the scenario:
  • There were 2 gateways which have per-gateway-dhcp applied, say, T-1, T-2.
  • A segment using gateway-dhcp connected to T-1 at first.
  • The segment was moved from T-1 to T-2 later.
  • No other change to this segment before the upgrade happened.
  • The setup was upgraded to a newer release.
• Because of this issue:
  • The segment realization state remains in an “IN-PROGRESS” state.
  • The connected gateway realization state remains in a “FAILED” state.
  • Any changes made to the segment DHCP configuration fails to be realized.

Environment

VMware NSX 3.2.4 being upgraded to 4.2.3

Cause

This is caused by the inconsistent properties in the Segment introduced by the data-migration during upgrade, i.e. the migration bumped up the the Segment revision number but did not update the previous-state correctly.

Resolution

This issue is fixed in NSX 9.0 and NSX 4.2.3.3 EP

For a workaround in your current NSX version, please collect below data and contact Broadcom Support as workaround requires NSX Corfu database changes:

1. Collect GPRR corfu table from NSX Manager CLI (logged in as root):

/opt/vmware/bin/corfu_tool_runner.py --tool corfu-editor -n nsx -o showTable -t GenericPolicyRealizedResource > gprr.log

     2. Collect DHCP Server List using below API call from NSX Manager CLI (logged in as root):

curl -k -i --basic -u 'username':'password' 'https://<IP or localhost>/api/v1/dhcp/servers'