Re: Add -c to rsync commands on SR tutorial wiki page

From: Stephen Frost <sfrost(at)snowman(dot)net>
To: Jim Nasby <Jim(dot)Nasby(at)BlueTreble(dot)com>
Cc: Pg Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Add -c to rsync commands on SR tutorial wiki page
Date: 2016-08-18 18:31:58
Message-ID: 20160818183158.GQ4028@tamriel.snowman.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

* Jim Nasby (Jim(dot)Nasby(at)BlueTreble(dot)com) wrote:
> On 8/17/16 9:46 PM, Stephen Frost wrote:
> >* Jim Nasby (Jim(dot)Nasby(at)BlueTreble(dot)com) wrote:
> >>> https://wiki.postgresql.org/wiki/Binary_Replication_Tutorial does
> >>> not specify -c for any of the rsync commands. That's maybe safe for
> >>> WAL, but I don't think it's safe for any of the other uses, right?
> >>> I'd like someone to confirm before I just change the page... my
> >>> intention is to just stick -c in all the commands.
> >-c is only relevant when you are doing an incremental copy, but on a
> >quick look, all those rsync commands appear to be doing full copies?
> >
> >You would want -c if you were taking a backup and then doing an update
> >of it using rsync. or something along those lines, as you can't really
> >trust rsync's time/size based comparison as it only has a 1 second level
> >granularity.
>
> I don't think it's any great leap for someone to think they can use
> those commands incrementally. It's certainly one of the first things
> you think of when using rsync. AFAIK there's no downside at all to
> using -c when it is a brand new copy, so I'm thinking we should just
> put it in there, especially considering what the potential downside
> is.

To have proper incremental backups done requires a lot more than just
throwing "-c" into the rsyncs. For my 2c, I'm at the point where I'd
prefer we discourage people from using rsync, cp, or generally try to
set up their own hand-rolled backup system with PG. Those examples
simply encourage poor setups that risk losing data and introducing
corruption.

Thanks!

Stephen

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Jim Nasby 2016-08-18 18:35:23 Re: [PATCH] bigint txids vs 'xid' type, new txid_recent(bigint) => xid
Previous Message Jim Nasby 2016-08-18 18:30:57 Re: synchronous_commit = remote_flush