This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.
Hi Justin, > We are seeing some unexpected behavior in the interaction of server and > client LDMs, I hope you can shed some light on this. > > We are running LDM on two separate machines and consider only one to be > primary at a time. We have a hostname, "madisprime", pointing to the IP > address of whichever of the two machines we consider primary. There are three > downstream LDMs requesting data from "madisprime" in their ldmd.conf. > > Today we made a failover of the primary LDM by reassigning the IP address of > "madisprime; we did not stop any of the LDMs. Our expectation was that the > downstream LDMs would begin pulling from the newly assigned "madisprime" LDM > fairly quickly but after several hours only one feed type has switched over. > The other 5 feed types are still being pulled by the former "madisprime" > system. As long as the TCP connection between an upstream LDM and a downstream LDM stays up, the downstream LDM will continue to get data from the upstream LDM. > Do you have any suggestions for how we can reduce the time it takes for the > feeds to check which address they should be pulling from? We believe > restarting either the former primary or the downstream LDMs would cause the > connections to be reestablished to the new "madisprime" system. > > We are hoping that there is a more automatic method. The only way to do what you want is to either restart the downstream LDM-s or stop or restart the upstream LDM on Madisprime. Stopping or restarting the LDM on Madisprime could be done automatically by a script. Alternatively, you could have the downstream LDM-s request the same data simultaneously from both the primary and secondary LDM servers. Then it wouldn't matter if one of them went offline. We do that here and most of our Internet sites do the same. The LDM protocol is designed for this. > Thanks, > > Justin Cooke > NCEP Central Operations Regards, Steve Emmerson Ticket Details =================== Ticket ID: NZT-443019 Department: Support LDM Priority: Normal Status: Closed