Tinderbox spam

[Alexandru Popa]

--- On Mon, 9/24/12, Dag-Erling Smørgrav <des at des.no> wrote:

> From: Dag-Erling Smørgrav <des at des.no>
> Subject: Re: Tinderbox spam
> To: "Chris Rees" <utisoft at gmail.com>
> Cc: stable at freebsd.org
> Date: Monday, September 24, 2012, 5:51 PM
> Chris Rees <utisoft at gmail.com>
> writes:
> > "Dag-Erling Smørgrav" <des at des.no> writes:
> > > Chris Rees <utisoft at gmail.com>
> writes:
> > > > Have you tried using http:// ?
> > > Both should work, but svn is significantly
> faster.
> > Yes, that's why I tried it instead, but my point is
> that you may need
> > to sleep a bit between tries; if svn is faster, you're
> more likely to
> > hit any rate limit.
> 
> It sleeps for 30, then 60, then 90 seconds and gives up
> after the fourth
> attempt.

What about rate limiting between parallel checkouts for builds for different architectures?  Are those staggered, or will svn.freebsd.org be seeing 6-7 different svn connections starting at the same time, when the builds start?

The logs seem to show that all attempted builds start at the exact same time.

Examples (snipped from 3 different mails):
TB --- 2012-09-25 02:30:00 - starting RELENG_7 tinderbox run for i386/i386
TB --- 2012-09-25 02:30:00 - starting RELENG_7 tinderbox run for ia64/ia64
TB --- 2012-09-25 02:30:00 - starting RELENG_7 tinderbox run for amd64/amd64



> DES
> -- 
> Dag-Erling Smørgrav - des at des.no

Regards,
Alex