be me, paged at 4am half the users cannot reach the new service the other half can nothing wrong in the logs check the record: correct check from the office: wrong IP check from home: correct IP find the record's TTL: 86400 we moved the service yesterday did not lower the TTL beforehand some resolvers will keep the old answer for up to a day
There is no fix except waiting, or poking each affected resolver. The one thing I could have done was lower the TTL a day before the move. I knew that. I wrote it in a ticket in 2021.