SLAC logo

Packet loss on Stanford-Pac Bell DSL links - Feb 2001 Network logo

Les Cottrell. Page created: February 26, 2001.

Central Computer Access | Computer Networking | Network Group | ICFA-NTF Monitoring
SLAC Welcome
Highlighted Home
Detailed Home
Search
Phonebook

Introduction

Teresa Downey reported the following on Sunday 25, 2001 at 8:28pm:
I cannot do anything from home tonight. The worst it has ever been.
I can only type a few characters at a time. Then must wait up to a minute
or so for it to catch up.
At the same period Gary Buhrmaster and Robert Cowles both using the Pac bell DSL (not through Stanford) reported no problems getting to SLAC. A traceroute from Bob Cowles' home machine to SLAC shows the route through Pac Bell.

Randy Melen, also connected via Stanford U/Pac Bell, reported ping losses at 11:00pm 2/25/01, both from SLAC and Stanford U to his home machine. Randy also noted on 2/26/01 at 8:50am:

The problem is still there.  It is almost a pattern of 5
successful packets, 5 dropped packets, 5 successful packets, 5 dropped
packets, ...

Gary Buhrmaster around 9:30pm 2/25/01, noted that the problems appear to be in the Stanford network and noted that:

Accessing some Stanford
web sites is very very slow (this doesn't use
SLAC resources in any way). 
If you go to http://monitoring.stanford.edu you
will see some systems "partially operational"
and some "not operational".  I presume this is
not supposed to be that way.

Routing

A traceroute from SLAC to Teresa's home machine shows the normal route with no long delays, but possible packet loss.

A traceroute from Les Cottrell's home machine (atlas w.x.y.z ) shows a very short route, but with some very long delays.

Pingroute

Pingroutes to Les Cottrell's home machine over VPN and to his DSL router (not over VPN) indicate the packet loss starts as one gets to the Stanford campus network.

Resolution

The problem was reported to Stanfor University network operations center by Les Cottrell at about 8:30am 2/26/01. At about 11:06 we received the following email from Wayne Sung of campus:
Pass this along to whomever you think - we somehow wound up with two
routers advertising the same ospf id, causing no end of consternation.
The observation of five good five bad was correct - we were recalculating
every ten seconds.

Page owner: Les Cottrell