Accessing Toronto Public Library's Website with Linux

D. Hugh Redelmeier hugh-pmF8o41NoarQT0dZR+AlfA at public.gmane.org
Sun Dec 21 16:32:17 UTC 2008


| From: Giles Orr <gilesorr-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org>

| http://support.microsoft.com/kb/934430/en-us

Thanks for that reference.  Interesting.

| Although Microsoft's web page claims "This issue does not occur when
| you try to make an HTTP connection," we've had reports of the listed
| fix helping people out.

Also interesting to hear.

| Does anyone know how to determine (from where we stand) what device -
| make and model - is causing the problem?  I believe we've let the
| network people know there is a problem, but it hasn't been fixed yet:
| perhaps if I can tell them precisely what device and how to fix it
| they would move on it.

If you can do a tcpdump on either side of each device, then you should
be able to see the option being mangled.

Here's an example of tcpdump output of such a packet, before mangling:
11:43:33.881345 IP xxxx.xxx.com.50058 > 192.30.226.5.http: S 3755011948:3755011948(0) win 5840 <mss 1460,sackOK,timestamp 59175877 0,nop,wscale 7>
Notice the "wscale 7".

I have no idea if this would work: traceroute works by setting
hop-counts so that messages get bounced at different distances from
the tracrouting node.  The resulting ICMP messages might contain
enough of the originating packet to see if an option is changed.  I
wonder if one could create a traceroute-like program that uses
suitably crafted SYN packets to probe for this deffect along a whole
path.

In any case, it looks to me as if TPL doesn't return the ICMPs
necessary:

10  core4-toronto63_POS12-0-0.net.bell.ca (64.230.160.125)  29.155 ms  28.905 ms  19.895 ms
11  core2-toronto63_POS0-2.net.bell.ca (64.230.242.106)  22.169 ms  25.951 ms  23.427 ms
12  dis33-toronto63_POS5-0.net.bell.ca (64.230.229.10)  23.889 ms  21.750 ms  26.669 ms
13  * * *
14  * * *
15  * * *
16  * * *
17  * * *

| P.S. I'm not at Lillian H. Smith anymore.

Too bad.  Handy to TLUG meetings.
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists





More information about the Legacy mailing list