Closed
Description
Hello !
I have some strange behavior of sequential pg_probackup executions.
postgres@server:~/backup-tools# ls /backup/postgresql/backup/wal/ | grep exampledb
exampledb
postgres@server:~/backup-tools# ls /backup/postgresql/backup/backups/exampledb
ls: cannot access '/backup/postgresql/backup/backups/exampledb': No such file or directory
postgres@server:~/backup-tools# pg_probackup-11 add-instance -B /backup/postgresql/backup -D /data/postgresql --instance exampledb --remote-host postgres.example.com --remote-user postgres
ERROR: arclog_path '/backup/postgresql/backup/wal/exampledb' already exists
postgres@server:~/backup-tools# ls /backup/postgresql/backup/backups/exampledb
postgres@server:~/backup-tools#
postgres@server:~/backup-tools# pg_probackup-11 add-instance -B /backup/postgresql/backup -D /data/postgresql --instance exampledb --remote-host postgres.example.com --remote-user postgres
ERROR: instance '/backup/postgresql/backup/backups/exampledb' already exists
So I think this behavior is not idempotent. When we try to execute pg_probackup several times one after another pg_probackup not only raises different error messages but also create inventory in broken state.
I think that before making any changes to file system pg_probackup should validate all conditions that possibly can prevent to create instance inventory successfuly.
Thank you in advance.