Hi Tinus, -(long version discarded, using short version)-
> so getting back to my original idea, with added advice: > > -change TTL lower it to some mins, not from 2d to 1d =) A TTL of 600 or less should do for the time of the transition. > -(wait > 2* original TTL) > -move over data (assuming data is not changed during this time) having it replicated inside the backend services would be preferred, but manually hard syncing and making sure that the old server isn't changed anymore is also a valid alternative. > -change zone files to reflect new situation & redirect any web/pop/smtp/ftp > connections you don't really need to forward traffic except for the 10min timeout of the 600s TTL in the DNS. So very very likely you don't even need the forwarding if the above was prepared well. Changing the old dns to be a slave of the new dns will save you from forgettting to copy over changes in the new DNS server. Shutting down services for these 10 min at the old server will also make sure that no new data is being uploaded and thus making the new host out-of-sync. (usually a maintenance downtime of that length should be ok unless you run HA services, but then you can just add replication in the services layer) > -change host record of main DNS at the registrar > -wait a bit & take OLD down. > > that right? Yes, except for the comments above. -- Best regards, Kilian
signature.asc
Description: Dies ist ein digital signierter Nachrichtenteil
