Greetings,
On Sun, 28 Jul 2019, Brian Kantor via 44Net wrote:
As I said, time will fix that - it's a question of
when the old
address expires out of DNS caches and the new one is learned.
It *should* be a multi-step process:
1) The Hostmaster sets the TTL cache time setting for the impacted
Resource Record in DNS to less than one hour.
2) The Sysadmin waits until the TTL cache time has gone by. So if it
was set to 3 days, s/he *WAITS* 3 days *before* performing the IP
address cutover. This ASSURES that no one will cache the old IP
addres for more than 1 hour.
3) The SysAdmin executes the IP re-addressing of the desired server.
The Hostmaster also updates the new IP address into the DNS resource
record, but *LEAVES* the TTL cache time set to a SHORT period of
time (1 hour) until users report NO problems with the server at the
new address. In this way *IF* you have to back out the change, and
switch back to the original address, no one will be caching the NEW
address for more than an hour. Keep the impact low....
4) Once it is verified that NO ONE is having problems with the new IP
address and the sevice is fully functional at its new address, the
Hostmaster will up the TTL cache time back to the original setting
(3 days, or whatever).
*NEVER* just switch the IP address of an important server and then
*blame* DNS Cache for unreachability problems, when in reality it was just
poor execution :(
Sure, email from
MAILMAN.AMPR.ORG is *not* life saving nor is it going
to affect anyone's lives, but practicing good Sysadmin skills is "A Good
Thing"(tm) :)
Thanks for the work you do,
--- Jay WB8TKL