k1det2 which worked as a product version of the bruco server was swapped to a test server because the product server had a similar problem (related to leap seconds) with klog#32382.
OS of the product server was already expired. So OS of the test server is different from the old one.
But the test server has been running in parallel with the production server for the recent several months, and there should be no problem.

-----
Detailed procedures
- Unmount a data region on the both the product and the test servers
- Swap IP addresses of these two servers.
- Restart nfs-kernel-server on k1nds2 (I couldn't understand why it's required but I couldn't mount the data region from the bruco server without restarting and I lost lots of time to solve it.)
- Remount the data region
- Update hostkey of the bruco server which is stored on all client workstations