Hi,
Sorry last posting was accidently sent , incomplete,plese ignore.
We are trying to setup PITR system based on documentation
http://www.postgresql.org/docs/8.1/interactive/backup-online.html We have 3 questions regarding
1. Selection of archived WAL records that can be safely removed.
2. Knowing the relevant WAL record just after pg_stop_backup()
3. Recommended tool for taking file system level backup.
Elaboration:
we want to remove archived WAL log files which are no longer needed
so that the size of wal_archive folder is under control.
The DOCs say that filenames numerically less than the WAL record that pg_stop_backup()
suggests can be removed. Will an alphabetical sorting be different from numerically sorted
filename ? Sorting numerically is not easy as the filenames look like HEX numbers
ut they are are so huge that they cant be stored in normal integers (32bits).
Q2. We are attempting to automate the process of taking base backup and removal
of the unneeded WAL files. Is there any reliable way of knowing the WAL file X from
inside the shell script , such that files prior to X can be removed ?
from a shell script we issue pg_stop_backup() by psql -c , it returns something
which does not looks like a WAL file.
Q3. tar exits with non zero status for the same reasons as mentioned in docs
is there any better archiving tool for this purpose ? can we use cp -a ?
Regds
Rajesh Kumar Mallah.