Re: Updated backup APIs for non-exclusive backups - Mailing list pgsql-hackers

On Tue, Apr 5, 2016 at 5:35 PM, Magnus Hagander <magnus@hagander.net> wrote:

On Mon, Apr 4, 2016 at 3:15 PM, Amit Kapila <amit.kapila16@gmail.com> wrote:

On Mon, Apr 4, 2016 at 4:31 PM, Magnus Hagander <magnus@hagander.net> wrote:

On Fri, Apr 1, 2016 at 6:47 AM, Amit Kapila <amit.kapila16@gmail.com> wrote:

 

Also, I think below part of documentation for pg_start_backup() needs to be modified:

<para>

    <function>pg_start_backup</> accepts an

    arbitrary user-defined label for the backup.  (Typically this would be

    the name under which the backup dump file will be stored.)  The function

    writes a backup label file (<filename>backup_label</>) and, if there

    are any links in the <filename>pg_tblspc/</> directory, a tablespace map

    file (<filename>tablespace_map</>) into the database cluster's data

    directory, performs a checkpoint, and then returns the backup's starting

    transaction log location as text.  The user can ignore this result value,

    but it is provided in case it is useful.

That one definitely needs to be fixed, as it's part of the reference. Good spot.

 

Similarly, there is a description for pg_stop_backup which needs to be modified.

That's the one you're referring to in your first commend above, is it not? Or is there one more that you mean?

I am referring to below part of docs in func.sgml

<para>

    <function>pg_stop_backup</> removes the label file and, if it exists,

    the <filename>tablespace_map</> file created by

    <function>pg_start_backup</>, and creates a backup history file in

    the transaction log archive area.  The history file includes the label given to

    <function>pg_start_backup</>, the starting and ending transaction log locations for

    the backup, and the starting and ending times of the backup.  The return

    value is the backup's ending transaction log location (which again

    can be ignored).  After recording the ending location, the current

    transaction log insertion

    point is automatically advanced to the next transaction log file, so that the

    ending transaction log file can be archived immediately to complete the backup.

 </para>

Ah, gotcha. Clearly I wasn't paying attention properly.

PFA a better one (I think), also with the rename and added comments that David was asking for.

This version looks good.  Apart from review, I have done some tests (including regression test), everything seems to be fine.