Re: Backup history file should be replicated in Streaming Replication?

Tom Lane wrote:
> Simon Riggs <simon(at)2ndQuadrant(dot)com> writes:
>> The backup history file is a slightly bit quirky way of doing things and
>> was designed when the transfer mechanism was file-based.
>
>> Why don't we just write a new xlog record that contains the information
>> we need?
>
> Certainly not. The history file is, in the restore-from-archive case,
> needed to *find* the xlog data.

Hmm, not really. The backup_label file tells where the checkpoint record
is. And that is still needed. AFAICS the backup history file is only
needed to determine the point where the backup was completed, ie. the
minimum safe stopping point for WAL replay.

I think we could get away without the backup history file altogether.
It's kind of nice to have them in the archive directory, for the DBA, to
easily see the locations where base backups were taken. But if we write
the backup stop location in WAL, the system doesn't really need the
history file for anything.

--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com