On Fri, Nov 20, 2015 at 12:53 PM, Philip Martin <philip.martin_at_wandisco.com>
> Mark Phippard <markphip_at_gmail.com> writes:
> > I should have added that the part that I would question here is the value
> > and importance of the cache. Which is the root of this. I just do not
> > it. I guess you (Philip) do else you would not be looking into this.
> > majority of our users, as my educated guess only, are probably using
> > server with prefork MPM.
> Apache prefork typically has low performance because the maximum number
> of processes is high and that limits the size of the FSFS cache. Use a
> threaded MPM, reduce the number of processes and increase the cache size
> to get better performance.
Does SVN work reliably with the threaded MPM? I know it works but recall
there were issues with pool lifetimes etc. since the top-level pools have
such a long lifetime. Maybe there were other issues in the past. I am
sure things are better than they were 7-8 years ago but I've always just
stuck with prefork as the safe/predictable option. It is nice when you
have those huge REPORT requests that the process just goes away when it is
> I can use my local mirror of Subversion to demonstrate the effect of the
> FSFS cache. I setup localhost svn+ssh and svn-with-SASL access, both
> require authentication and both do encryption.
> Running 'svnbench null-log' on trunk takes:
I am just more interested to what we would see in the wild. If the ASF
ever upgrades the server to 1.9 do you think we are going to see noticeable
performance boosts from the cache? I don't.
All of my SVN servers are Apache, and most of them serve hundreds if not
thousands of repositories. I just do not see a cache yielding much
Anyway, I am not trying to hijack your thread. It just seems like the only
people that will really benefit from these caches are those in a controlled
lab environment. I am just doubting that svn+ssh:// today pays any more of
a penalty than the typical Apache server does. So if the goal is to move
to SSH and certs it does not seem like this should be a real blocker.
Received on 2015-11-20 20:14:12 CET