Re: BUG #15335: Documentation is wrong about archive_command and existing files

Greetings,

* David Steele (david(at)pgmasters(dot)net) wrote:
> On 8/18/18 8:09 PM, Jeff Janes wrote:
> > I think it might be a good idea to propose use of a canned solution, but
> > I don't know how the community would feel about mentioning specific
> > projects by name.  And if we didn't mention any by name, I think it
> > would be pretty awkward to advise to give.
> There is a wiki page that lists the major backup solutions:
>
> https://wiki.postgresql.org/wiki/Binary_Replication_Tools

Interesting, though seems like there should be a distinction made
between "backup" tools and "replication" tools, they're certainly
different things.

> But I don't think we ever reference the wiki from the core user
> documentation.

Very rarely.

> We could at least list all the things that a good archive command should
> do and point out that the example in the docs doesn't do them and that
> it is intended *only* as an example.

This sounds like a good idea to me, in general. I suggest we qualify
the command shown further and say it's only provided as an illustration
of how to set an archive_command and that it isn't an example. Perhaps
we should even remove the shell-script bits and instead just have:

archive_command = 'archivecmd /mnt/server/archivedir/%f %p'

As long as we have something there that looks like valid shell script
and which doesn't obviously fail, people are likely going to continue to
use it.

Thanks!

Stephen