but deploy error already destroy the machine.
so the documents solution is useless.
next ‘flyctl deploy’ create new machine.
but because it is not set to ‘–restart always’, error occur again.
any prior to error configuration is not possible?
This issue might not be related to the restart policy. Were there any other error messages before or after the restart policy error? Try posting all the output from the failed fly deploy command, as well as the logs from that time period.