I am having a problem reaching what should be an easily reachable
server going through 4 ISPs. I have no problems pinging or
tracerouting to other hosts on the Internet. The target server is
reachable from just about anywhere except the DSL connection I'm going
through. The route goes through four ISPs: Xspedius, AT&T, Qwest, and
Speakeasy. Each one either states the routing problem is not on their
network or their noc is non-responsive. This has got to be a routing
or subnet mask problem with one of them.

I'm trying to reach 66.93.249.81 (a Speakeasy IP). The trace gets to
the speakeasy network and then dies. The last two hosts in the trace
below (13 and 14) are un-pingable as well:

03/20/06 14:31:35 Fast traceroute mia-c2.0spam.net
Trace mia-c2.0spam.net (66.93.249.81) ...
1 192.168.2.1 1ms 1ms 1ms TTL: 0 (No rDNS)
2 165.236.196.1 9ms 11ms 8ms TTL: 0 (No rDNS)
3 No Response * * *
4 199.227.6.81 16ms 11ms 19ms TTL: 0 (No rDNS)
5 216.85.80.5 33ms 24ms 82ms TTL: 0 (No rDNS)
6 12.119.94.17 44ms 41ms 39ms TTL: 0
(xspedius.fldfl.ip.att.net probable bogus rDNS: No DNS)
7 12.123.33.62 77ms 45ms 50ms TTL: 0 (No rDNS)
8 12.122.12.121 46ms 35ms 52ms TTL: 0
(tbr1-cl1474.attga.ip.att.net probable bogus rDNS: No DNS)
9 12.123.20.201 39ms 54ms 56ms TTL: 0
(br1-a340s8.attga.ip.att.net probable bogus rDNS: No DNS)
10 192.205.33.90 41ms 48ms 48ms TTL: 0 (No rDNS)
11 205.171.21.105 50ms 40ms 34ms TTL: 0
(atl-core-01.inet.qwest.net fraudulent rDNS)
12 205.171.221.18 44ms 51ms 56ms TTL: 0
(atx-edge-01.inet.qwest.net fraudulent rDNS)
13 208.45.200.102 39ms 40ms 51ms TTL: 0 (No rDNS)
14 69.17.82.122 44ms 48ms 43ms TTL: 0
(220.fe-2-3.er1.atl1.speakeasy.net probable bogus rDNS: No DNS)
15 No Response * * *
16 No Response * * *

Similar (yet not exact) routes from public looking glass servers have
no problem. The target server is VERY reachable.

Any suggestions as to where the routing problem is, and once
identified, who to contact so that something will actually get fixed.
I've worked the issue from the Speakeasy and the Xspedius ends, but no
one wants to own this problem. Extremely frustrating.

Thanks in advance. You can email me directly: bill @ 0spam.net

bill@0spam.net writes:
> I am having a problem reaching what should be an easily reachable
> server going through 4 ISPs. I have no problems pinging or
> tracerouting to other hosts on the Internet. The target server is
> reachable from just about anywhere except the DSL connection I'm going
> through. The route goes through four ISPs: Xspedius, AT&T, Qwest, and

Are you _sure_ it's a routing problem? If it is, then looks like it's
fairly clearly within speakeasy.net, as the last working hop is there
as are all of the hops past that point.

There's another possibility, though. Having the site be unpingable
and not appear on traceroute implies that ICMP messages aren't getting
through. Some "security expert" may have installed a filter on that
path that discards ICMP.

DSL lines often use PPPoE, which has an unusual MTU cap of 1492. In
order to contact most web sites on the net, you then may need working
path MTU discovery, which is in turn thoroughly broken when these
"experts" filter away ICMP.

The symptom for this would be that small-sized TCP packets get through
just fine, but the largest ones never do, and connections just time
out.

Is that what you're seeing?

--
James Carlson, KISS Network <james.d.carlson@sun.com>
Sun Microsystems / 1 Network Drive 71.232W Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677

1. Traceroutes and pings to the rest of the world work correctly, so I
don't think the problem is with the DSL provider.

2. Going from the other direction, I can ping 69.17.82.122 and
208.45.200.102. I can also ping them from just about any other
network. So I don't think that ping has been disabled on either
router.

3. I have seen problems in the past where an improper (too wide open of
a) netmask led to one-way routing problems.

4. The return traceroute uses asymmetrical routing supposedly - it
takes a very different return path:

# traceroute 165.236.196.1
traceroute: Warning: Multiple interfaces found; using 66.93.249.81 @
hme0
traceroute to 165.236.196.1 (165.236.196.1), 30 hops max, 40 byte
packets
1 dsl093-249-065.mia1.dsl.speakeasy.net (66.93.249.65) 0.705 ms
0.593 ms 0.487 ms
2 dsl092-214-001.atl1.dsl.speakeasy.net (66.92.214.1) 45.466 ms
39.226 ms 39.567 ms
3 220.ge-0-1-0.cr2.atl1.speakeasy.net (69.17.82.121) 39.361 ms
54.474 ms 64.200 ms
4 208.45.200.101 (208.45.200.101) 55.452 ms 62.003 ms 55.483 ms
5 205.171.221.13 (205.171.221.13) 63.955 ms 27.182 ms 40.397 ms
6 205.171.21.169 (205.171.21.169) 39.510 ms 39.607 ms 35.393 ms
7 0.so-3-0-0.BR1.ATL4.ALTER.NET (204.255.169.81) 33.017 ms 46.181
ms 39.552 ms
8 0.so-2-2-0.XL2.ATL4.ALTER.NET (152.63.86.174) 40.130 ms 54.755 ms
55.546 ms
9 0.so-3-0-0.CL2.JAX1.ALTER.NET (152.63.83.157) 62.869 ms 61.605 ms
63.073 ms
10 251.ATM6-0.GW5.JAX1.ALTER.NET (152.63.84.237) 63.028 ms 63.198 ms
87.193 ms
11 xspedius-jax-gw.customer.alter.net (65.195.232.254) 62.994 ms
124.978 ms 55.549 ms
12 ftldfl-01-so-2-0-0.xspedius.net (206.222.122.18) 71.659 ms 79.111
ms 78.778 ms
13 216.85.80.6 (216.85.80.6) 78.698 ms 70.297 ms 78.598 ms
14 199.227.6.82 (199.227.6.82) 79.002 ms 61.832 ms 63.063 ms
15 165.236.196.1 (165.236.196.1) 63.070 ms 65.742 ms *

5. Note that the 205.171.221 qwest node routes to the 205.171.21 qwest
node and then routes to 204.255.169 (alter.net) instead of 192.205.33
(att.net).

6. I'm pretty sure the problem is with 69.17.82.122 and 208.45.200.102.
Speakeasy can't test a route through 208.45.200.102 and qwest seems to
be unreachable - none of their public looking glasses route through
208.45.200.102.

In article <xoav1wwwszyf.fsf@sun.com>,
James Carlson <james.d.carlson@sun.com> wrote:

> bill@0spam.net writes:
> > I am having a problem reaching what should be an easily reachable
> > server going through 4 ISPs. I have no problems pinging or
> > tracerouting to other hosts on the Internet. The target server is
> > reachable from just about anywhere except the DSL connection I'm going
> > through. The route goes through four ISPs: Xspedius, AT&T, Qwest, and
>
> Are you _sure_ it's a routing problem? If it is, then looks like it's
> fairly clearly within speakeasy.net, as the last working hop is there
> as are all of the hops past that point.

If traceroutes from other places to the destination host make it
successfully through speakeasy.net, it's likely to be a problem with
speakeasy's route back to the sending address.

The best way to troubleshoot problems like this is by tracing in both
directions. You need to start a trace from somewhere on the destination
network, and see how far it goes. If Speakeasy has a "looking glass" it
may be able to show you their routes for your address.

--
Barry Margolin, barmar@alum.mit.edu
Arlington, MA
*** PLEASE post questions in newsgroups, not directly to me ***
*** PLEASE don't copy me on replies, I'll read them in the group ***