Hello
I am testing the concept of this extension, manually.
On my local database, if I modify a table and then make a copy of pgdata/base/XXX, the copy doesn't contain the modified data.
I need to call pg_backup_start('test', True) and keep the session open during the copy to ensure the data is flushed into pgdata/base/XXX directory.
It also seems that using subvolumes is not required.
This command works well:
cp -ra --reflink=always pgdata/base/159591 pgdata/base/111111
I'm not sure if this technique can be used for production software.
PostgreSQL documentation contains a lot of warnings related to filesystem copies: https://www.postgresql.org/docs/current/backup-file.html
Hello
I am testing the concept of this extension, manually.
On my local database, if I modify a table and then make a copy of pgdata/base/XXX, the copy doesn't contain the modified data.
I need to call
pg_backup_start('test', True)and keep the session open during the copy to ensure the data is flushed into pgdata/base/XXX directory.It also seems that using subvolumes is not required.
This command works well:
cp -ra --reflink=always pgdata/base/159591 pgdata/base/111111I'm not sure if this technique can be used for production software.
PostgreSQL documentation contains a lot of warnings related to filesystem copies: https://www.postgresql.org/docs/current/backup-file.html