Currently, pgautoupgrade has support for upgrading versions as far back as 9.6. PostgreSQL 9.6 hasn't had a release since late 2021, when it lost active support.
Dropping support for very old versions keeps the images smaller (currently compressed images are ~1.6x than standard postgres builds), and likely improves build time and maintenance. Therefore, I'd suggest dropping support for versions older than 12.
Keeping EoL versions likely makes sense, but perhaps only the oldest version? If users need to upgrade their 9.6 DB, it's not unreasonable for them to have to do it in 2 jumps rather than 1.
Currently,
pgautoupgradehas support for upgrading versions as far back as 9.6. PostgreSQL 9.6 hasn't had a release since late 2021, when it lost active support.Dropping support for very old versions keeps the images smaller (currently compressed images are ~1.6x than standard postgres builds), and likely improves build time and maintenance. Therefore, I'd suggest dropping support for versions older than 12.
Keeping EoL versions likely makes sense, but perhaps only the oldest version? If users need to upgrade their 9.6 DB, it's not unreasonable for them to have to do it in 2 jumps rather than 1.