Re: Burning a .raw file...
On Fri, 10 May 2002, Attila Nagy wrote:
> > > That machine is already dying. I would need some slot 1 PIII processors to
> > > speed that up, but they are very hard to find...
> > Well, in an effort to get a bit less load on it, I'm offering
> > push-triggered mirroring of the directory:
> > ftp.fsn.hu/pub/CDROM-Images/debian-unofficial/
> Wow, that's great!
Toffer at ftp.no.d.o that I already trigger for debian[-non-US] asked for
it. Since it takes so long to sync the dataset, without trigger you run
quite a large risk of trying to update during updates. So please, if you
have a mirror of the ftp.fsn.hu unofficial debian images and some
familarity with the debian push-sync system, consider sending me an email
to set it up.
A quick check reveales that:
Starting at Fri May 10 14:05:02 MET_DST 2002
Done at Sat May 11 05:14:46 MET_DST 2002
Perhaps I should move it up earlier in the day, that would make it easier
for my server (which has heavy load from about midnight to 07). But then,
even at the highest load it should be fast enoguh to keep up.
> We are before a major reconstruction of the site, so I hope in the future
> we can handle the load better (currently we are capped at 3000 concurrent
> FTP clients :).
Well, either something broke or you are having some kind of scheduled
maintenece right now, I can't get to it either.
> BTW, as turned out we have some problems with our ISP too, because their
> international pipes went saturated, so that's why we are so slow from
> outside Hungary.
Ok, that could explain the 11kB/s rates I see in my rsyncs. Or that could
be explained by the really stupid routing (going via .us).
> We are working on this, but first we have to "legalise" this service, so
> we decided to set up a foundation around that. This way we hope that we
> can get an agreement from our current ISP (with the service parameters
> laid down on paper) and we will try to get more hardware (AKA donations,
> so money will also count) to build a system where we can put several
> frontends to different ISPs in Hungary.
That sounds great! Hope you manage to find stuff and links to build an
even better server.
> The first, we will try have multiple 2.5 Gbps connections to GEANT, the
> European research network, but that race will be hard to win. :(
Yeah, you aren't really a university. But if you could do the multiple
frontend thing, you could perhaps put one on the research net? If it only
uses that capacity for geant/i2 routes, it shouldn't be much of a problem
except that of setting the links and routing up.
Also, perhpas you can point to the fact that other projects like this
exists on educational links like ftp.sunet.se (avg bw used 200-500Mbit/s)
run and paid for by the swedish university network itself.
> I hope that they can recognize the importance of this work, and they
> decide to support us.
> We'll see...
> > Hmm.. Regarding that, when is the best time for me to run that rsync
> > script? If I have others that mirror from me, it is probably good if I
> > manage to avoid mirroring just when the archive is being updated.
> Currently, the images are made daily on a separate machine and synced with
> the public archive nearly every week, when I think it is time to do that
> (I occassionally can test the images and if they are bad, I don't upload
> them, but that depends very much on my spare time).
> In the future I plan to do this differently, but that needs the current
> upgrade first.
Ok, I was asking because of what time of day I should do the syncs. I put
it at the arbitrary time 14:05 (local time here), because we have much
more night (mirroring) load than day (end-user) load. Please tell me if
you have any preferences.
To UNSUBSCRIBE, email to firstname.lastname@example.org
with a subject of "unsubscribe". Trouble? Contact email@example.com