rsync daemon performance
jamie at shareable.org
Tue Mar 9 12:20:32 MST 2010
Paul Slootman wrote:
> On Sat 06 Mar 2010, Wayne Davison wrote:
> > On Thu, Feb 25, 2010 at 5:20 PM, Tom Dickson <tdickson at finarta.net> wrote:
> > > Is it possible to get a "pool" of waiting daemons, similar to how apache
> > > runs?
> > No, there is no support for that in rsync at the moment. I don't think it
> > would be simple to add, but if someone wants to work on patch, I'd consider
> > adding it to rsync (whether as a feature or a patch would depend on the
> > change).
> I wonder what the benefit would be.
> Now I find it's useful that rsync parses the rsyncd.conf at connect
> time, not when the daemon is started.
Me too. But a multi-connection daemon could easily
stat("/etc/rsync.conf") on each new connection.
> The overhead of rsync starting a new process to handle a client is also
> a lot lower relative to the disk IO resulting from any transfer as well;
> especially compared to a typical apache situation where a page view may
> result is multiple parallel HTTP connections to the server.
I'm not sure about the overhead: if the data is already in cache, there
is no disk I/O, just system calls.
However, on Linux the time to start a new process is only a few
hundred microseconds, so I don't think it's a big deal unless you have
a large number of connections starting continuously, each doing only a
small bit of work.
More information about the rsync