NSX Federation: Global Manager manual switchover from Standby to Active timed out

Products

• VMware NSX

Issue/Introduction

In an NSX Federation environment, selecting the Standby cluster in Location Manager on the Global Manager and selecting Make Active will progress to 67% and then time out.

The error "Switching to standby timed out. Do you still want to continue?" is seen and options are presented to cancel, retry and force to active.

During some attempts, an error may be seen stating "Error: The policy object GlobalManager '(HOSTNAME)' used in this operation, has a different version (#) than the current system version (#). Fetch the latest copy of the object and retry operation. (Error code: 500071)"

Cause

Issues with NSX database content and database replication can cause this condition to occur, one of which is stale entries for clusters that no longer exist.

To identify whether this is the case, review log files with the name nsxplugin ... .log and look for replication status information near the phrase 'aggregated status response'.

For example, in /var/log/corfu-log-replication/nsxplugin.9999.2.log, information similar to this is found:

Returning aggregated status response 'correlation_id { ########-####-###### }
subscriber: LR
sync_status {
  standby_site_id: "11111111-####-####-111111111111"
  percentage_done: 100
  sync_type: DELTA_SYNC
...
}
sync_status {
  standby_site_id: "22222222-####-####-222222222222"
  percentage_done: 100
  sync_type: DELTA_SYNC
...
}

sync_status {
  standby_site_id: "33333333-####-####-333333333333"
  percentage_done: 100
  sync_type: DELTA_SYNC
...
}
}

In this example, there are three entries. If more than one entry is present, this is an indication that stale information is present in the NSX database.

Resolution

Contact support for issue confirmation and resolution. See Creating and managing Broadcom cases.

A future version of NSX will contain a software improvement to address this condition.