RE: FSFS packing: experiences?
From: Johan Corveleyn <johan.corveleyn_at_uz.kuleuven.ac.be>
Date: Thu, 18 Jun 2009 18:08:51 +0200
> I don't think so, I've not seen any performance improvements using it.
After some analysis of svn log, I think revs are certainly just as relevant as revprops. I've been meaning to start another (user or dev) thread about that (and I will), but here is some output of "rwsnoop" on our Solaris SVN server (1.5.4). We captured this while running "svn log https://svnserver/svn/trunk/file_with_lots_of_revisions".
-------- extract from rwsnoop
I guess that "svn log <path>" needs to read the relevant revs as well, just to determine which was the preceding revision that should be read next (information that's not present in the revprop files). Note that I didn't use "-v", so I didn't ask for the paths that were affected by the revisions (in that case, I could certainly understand the need to read the rev files).
Anyway, I've been meaning to discuss this, because I think there is a lot of room for improvement here, but now I just wanted to know if anyone has seen "real-world" improvement by packing. Has anyone ever compared timings of log and/or annotate pre- and post-packing? That would be really interesting ...
Johan
------------------------------------------------------
To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_subversion.tigris.org].
|
This is an archived mail posted to the Subversion Users mailing list.
This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.