On Tue, 15 Nov 2011, Pat Riehecky wrote:
> On 11/15/2011 12:49 AM, g wrote:
>> On 11/15/2011 05:56 AM, Jon Peatfield wrote:
>> > rsync://rsync.scientificlinux.org/ has been failing with:
>> >
>> > @ERROR: max connections (30) reached -- try again later
>> >
>> > are too many of us trying to fetch by rsync or should we have changed to
>> > a
>> > different server at some point?
>>
>> suggestions: stager cron times.
>
> I've increased the connection limit on the rsync system, and I would echo the
> comment above on varied mirror times. If you've been getting the connection
> limit error message, I'd suggest adding a random sleep before your rsync job.
Thanks, the errors seem to have gone away.
Just to be clear this is to update our local site mirror (serving about
600 machine). We only have one server connecting to you - making 4
connections one after the other as we pull over bits for sl3,sl4,sl5,sl6 -
we don't mirror the entire trees just the bits we think we need...
In fact it does already have a random delay in it though since that is
just the time taken to process various local repos it may not vary that
much on a typical day.
How large do you suggest making the random variation?
We currently run the scripts at about 03:37 (GMT0BST), so I suppose that
may correspond to a busy time elsewhere in the world - should we move it
to a different time?
> The increased connection count may lower performance slightly, however, it is
> not expected to be a significant loss in performance.
>
> Pat
-- Jon
--
/--------------------------------------------------------------------\
| "Computers are different from telephones. Computers do not ring." |
| -- A. Tanenbaum, "Computer Networks", p. 32 |
---------------------------------------------------------------------|
| Jon Peatfield, _Computer_ Officer, DAMTP, University of Cambridge |
| Mail: [log in to unmask] Web: http://www.damtp.cam.ac.uk/ |
\--------------------------------------------------------------------/
|
