On Sun, Dec 11, 2011 at 04:49:29PM +0100, Olaf van der Spek wrote:
On Mon, Dec 5, 2011 at 12:26 PM, Olaf van der Spek <ml@vdspek.org> wrote:
I assume we agree on the performance problems of the current tracker, right?
Lars?
In my eyes, the problem is not with the tracker. It's the underdimensioned set of servers we have. Rene probably has decent figures on what things are consuming the server resources, and might have a decent battle plan for migration to more servers. I don't see any point in jumping to conclusions about whether a piece of software should be replaced simply because the net load of all services is high. Everything on the svn.boost.org server is collectively slow to some degree, may it be web, trac, svn, regression tables, etc. A more productive avenue of investigation would be to find out whether services can be deployed over more machines, and whether that would improve things. -- Lars Viklund | zao@acc.umu.se