[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Using a backup server as local svn handler?

From: Pierre Fourès <pierre.foures_at_gmail.com>
Date: Thu, 22 Aug 2019 16:11:22 +0200


Le jeu. 22 août 2019 à 13:46, Andreas Stieger <Andreas.Stieger_at_gmx.de> a écrit :
> The only extra consideration that would need to be made is when using write-through proxying (SVNMasterURI). Merely noting for completeness that you would need an extra Location block in your httpd configuration. But your back-up configuration does not include that.
> http://svnbook.red-bean.com/nightly/en/svn.serverconfig.httpd.html#svn.serverconfig.httpd.extra.writethruproxy
> Andreas

Thanks for the detailed answer Andreas, you might have just pointed me
to what I was seeking to reach for my requirements of "quite high
availability" without a too complex setup. I doesn't need real and
strong high availability. I don't mind not to be able to commit
anymore if the master goes down, and while it's down. However, I
clearly would benefit to have the slave still be up and up-to-date
with the latest revision committed to the master in the event the
master would go down. For one point, the repos would be available in
read-only mode, but more over, if my understanding are good, depending
on my configuration, this would mean I could ensure not to loose a
single commit even if the master goes down. Is that last hypothesis
valid ?

In the documentation, I read "The extra bits on the end of each line
aren't necessary, but they're a sneaky way to allow the sync commands
to run in the background so that the Subversion client isn't left
waiting forever for the commit to finish.". Regarding my desire not to
loose any single commit, if I doesn't span the svnsync in the
background (by not using "&"), this should hang the client process
until the slave are synced, thus, when I commit, the commit would
return only when the slave is synced. This would slow up commits, for
sure, but if master and slave are connected to each other over a good
network, this shouldn't add a lot of lag and could be a trade off I
would be ready to take (for some repos) if this enable me to ensure
that when a commits returns, it has been propagated to at least two
(or more) physical servers.

I would really appreciate your view on the feasibility of this setup.

Best Regards,
Received on 2019-08-22 16:11:55 CEST

This is an archived mail posted to the Subversion Users mailing list.