Does Cache/Ensemble cache DNS lookups?
Hi
Some of our integrations use a system DSN to interact with our Data Warehouse (built on Microsoft SQL Server). We've recently migrated from SQL Server 2016 to 2022 and on the day of go-live we changed DNS cname records as a mechanism for redirecting everything to the new server. This worked great for users and reports, but for some reason we found that HealthShare clung onto the old server name even after several days.
We found when an ODBC connection references a DNS cname, the production hung onto and still queried the old server - we had to restart the production to resolve it.
When an inbound file service references a cname in a file path, the production hung onto the old server rather stubbornly. Restarting the production did not resolve this for us; we had to update the file path to the actual new server name, then set it back to the cname after which it has stayed pointing to the correct new server.
How does HealthShare cache/persist the servername from a DNS lookup, and is there a way to control its behaviour (maybe reduce ttl) or turn off caching altogether? There has to be a more elegant way of picking up a DNS change than the hacky methods we've found
Thanks
Failing somebody who has run into this before, I would suggest opening a WRC ticket so somebody from InterSystems can look at the relevant code and see what (if anything) can be done to avoid this behavior. Once you get that reply I'm sure there would be no problem sharing with the class here 😉