On 12/20/07, Bünzli Daniel <[EMAIL PROTECTED]> wrote:
>
> Le 20 déc. 07 à 17:31, Chris Nighswonger a écrit :
>
> > Next time from both locations do a traceroute from *nix teminal to
> > both problem sites.
>
> Since this is happening right at the moment, here you have :
>
> > > traceroute www.koha.org
> > traceroute to www.koha.org (66.249.19.54), 64 hops max, 40 byte
> > packets
> >  1  * * *
> >  2  1-48.2-85.cust.bluewin.ch (85.2.48.1)  16.947 ms  20.818 ms
> > 17.320 ms
> >  3  net481.bwrt2zhb.bluewin.ch (195.186.121.1)  26.200 ms  19.811
> > ms  18.686 ms
> >  4  po20.bwrt2inb.bluewin.ch (195.186.0.101)  15.428 ms  21.826 ms
> > 23.696 ms
> >  5  195.186.0.126 (195.186.0.126)  17.993 ms  18.317 ms  24.928 ms
> >  6  i79zhb-015-vla200.bb.ip-plus.net (138.187.129.62)  22.771 ms
> > 21.917 ms  18.081 ms
> >  7  i79tix-025-xxx1-1.bb.ip-plus.net (138.187.129.82)  15.307 ms
> > 19.798 ms  17.700 ms
> >  8  sl-gw10-zur-5-0-1.sprintlink.net (80.93.9.125)  21.013 ms
> > 14.940 ms  25.695 ms
> >  9  sl-bb20-zur-13-0.sprintlink.net (80.93.8.37)  26.198 ms  15.977
> > ms  21.069 ms
> > 10  sl-bb20-fra-11-1.sprintlink.net (213.206.129.155)  21.379 ms
> > 22.794 ms  26.077 ms
> > 11  sl-bb21-par-4-0.sprintlink.net (213.206.129.149)  31.866 ms
> > 31.188 ms  35.398 ms
> > 12  sl-bb24-nyc-9-0.sprintlink.net (144.232.9.166)  114.674 ms
> > 119.987 ms  116.966 ms

There appears to be a bottleneck between hop 11 and 12 on Sprint's
network. (Not uncommon on this provider from my experience with them.)
Looks like a long-haul link from Paris to New York my guess from the
naming conventions.

>
> > > traceroute www.kohadocs.org
> > traceroute to www.kohadocs.org (66.249.19.54), 64 hops max, 40 byte
> > packets
> >  1  * * *
> >  2  1-48.2-85.cust.bluewin.ch (85.2.48.1)  21.717 ms  14.797 ms
> > 16.971 ms
> >  3  net481.bwrt2zhb.bluewin.ch (195.186.121.1)  22.746 ms  17.641
> > ms  19.881 ms
> >  4  po20.bwrt2inb.bluewin.ch (195.186.0.101)  16.482 ms  24.118 ms
> > 23.033 ms
> >  5  195.186.0.126 (195.186.0.126)  30.118 ms  17.598 ms  26.106 ms
> >  6  i79zhb-015-vla200.bb.ip-plus.net (138.187.129.62)  25.305 ms
> > 18.867 ms  22.044 ms
> >  7  i79tix-025-xxx1-1.bb.ip-plus.net (138.187.129.82)  16.521 ms
> > 38.712 ms  16.316 ms
> >  8  sl-gw10-zur-5-0-1.sprintlink.net (80.93.9.125)  20.904 ms
> > 37.214 ms  18.249 ms
> >  9  sl-bb20-zur-13-0.sprintlink.net (80.93.8.37)  27.121 ms  17.717
> > ms  19.475 ms
> > 10  sl-bb20-fra-11-1.sprintlink.net (213.206.129.155)  26.492 ms
> > 21.837 ms  24.078 ms
> > 11  sl-bb21-par-4-0.sprintlink.net (213.206.129.149)  39.422 ms
> > 39.418 ms  49.981 ms
> > 12  sl-bb24-nyc-9-0.sprintlink.net (144.232.9.166)  190.203 ms
> > 154.597 ms  136.152 ms

Same link is bottlenecked here as well.

>
> but, (always so far) reachable :
>
> > > traceroute git.koha.org
> > traceroute to git.koha.org (64.79.208.60), 64 hops max, 40 byte
> > packets
> >  1  * * *
> >  2  1-48.2-85.cust.bluewin.ch (85.2.48.1)  15.283 ms  14.775 ms
> > 15.158 ms
> >  3  net481.bwrt2zhb.bluewin.ch (195.186.121.1)  16.576 ms  13.775
> > ms  15.185 ms
> >  4  po20.bwrt2inb.bluewin.ch (195.186.0.101)  15.237 ms  15.936 ms
> > 15.530 ms
> >  5  195.186.0.126 (195.186.0.126)  21.011 ms  14.527 ms  15.446 ms
> >  6  i00nye-005-pos12-0.bb.ip-plus.net (138.187.159.6)  111.069 ms
> > 112.085 ms  111.039 ms

Bottleneck between hops 5 and 6 here. Again, I'm guessing a long haul
link from Europe to New York based on the hostname of hop 6. hop 5
appears to be in Switzerland (.ch tld).

> > 16  gimli.metavore.com (64.79.208.60)  208.433 ms  207.933 ms
> > 206.965 ms

Notice here, the server hosting koha git did respond. However, with
RTT's like these there is a good chance some of the packets TTL's are
expiring before they reach one end or the other.

The server hosting the two websites may be behind a firewall or the
gateway may be setup to drop ICMP packets thus there is no final
response in the traceroutes for those sites.

Not much you or koha can do abou this. Probably the State side routers
are overloaded. Sprint has a bad habit of doing this in my experience.
There may also be TCP window scaling misconfigurations between the
endpoints on the trans-atlantic links which will cause these kinds of
issues as well.

Traceroutes from both of your two different physical locations would
probably confirm issues on the same endpoints.

Kind regards,

-- 
Chris Nighswonger
Faculty Member
Network & Systems Director
Foundations Bible College & Seminary
www.foundations.edu
www.fbcradio.org


_______________________________________________
Koha-devel mailing list
Koha-devel@nongnu.org
http://lists.nongnu.org/mailman/listinfo/koha-devel

Reply via email to