More Packet Loss in Leeds

chrisdev · 29-11-2004, 11:27

Hi,
Is anyone else seeing some unacceptable PL in Leeds this morning..?

Here's a snippet...

Reply from 158.43.128.1: bytes=32 time=10ms TTL=241
Request timed out.
Request timed out.
Reply from 158.43.128.1: bytes=32 time=20ms TTL=241
Reply from 158.43.128.1: bytes=32 time=10ms TTL=241
Request timed out.
Request timed out.
Reply from 158.43.128.1: bytes=32 time=20ms TTL=241
Reply from 158.43.128.1: bytes=32 time=20ms TTL=241
Reply from 158.43.128.1: bytes=32 time=20ms TTL=241
Reply from 158.43.128.1: bytes=32 time=20ms TTL=241
Ping statistics for 158.43.128.1:
Packets: Sent = 1004, Received = 583, Lost = 421 (41% loss)
Approximate round trip times in milli-seconds:
Minimum = 10ms, Maximum = 391ms, Average = 12ms

Dan Elwell's speed tester reports the same 41% loss to UK servers.

C:\>tracert 158.43.128.1
Tracing route to cache0002.ns.eu.uu.net [158.43.128.1]
over a maximum of 30 hops:
1 10 ms 10 ms 10 ms 10.63.192.1
2 10 ms 10 ms <10 ms leed-t2cam1-a-ge95.inet.ntl.com [80.0.54.13]
3 * * * Request timed out.
4 * 10 ms 10 ms lee-bb-a-so-120-0.inet.ntl.com [62.253.185.17]
5 * 10 ms 10 ms nth-bb-b-so-000-0.inet.ntl.com [62.253.185.101]
6 * * 10 ms pop-bb-a-so-100-0.inet.ntl.com [213.105.172.14]
7 * 40 ms 30 ms tele-ic-2-so-000-0.inet.ntl.com [62.253.185.86]
8 10 ms 20 ms 20 ms fe0.br3.lnd1.bb.uk.uu.net [195.66.226.16]
9 * 10 ms 20 ms srp4-0.cr1.lnd8.gbb.uk.uu.net [158.43.189.2]
10 * 10 ms 20 ms pos0-2.cr1.lnd4.gbb.uk.uu.net [158.43.254.246]
11 * * 20 ms pos6-0-0.hr1.lnd4.gbb.uk.uu.net [158.43.172.98]
12 10 ms * * fe0-0.lnd4eusoesr1.lnd.ops.eu.uu.net [158.43.254
.14]
13 20 ms 20 ms * cache0002.ns.eu.uu.net [158.43.128.1]
14 * * * Request timed out.
15 20 ms * 10 ms cache0002.ns.eu.uu.net [158.43.128.1]
Trace complete.

Northern-Monkey · 29-11-2004, 14:16

Yup, I have the same problem.

On my tracert after th 3rd hop (leed-t2core-a-ge-wan61.inet.ntl.com [80.0.241.33]) I get loads of packetloss, I think the router may be having a bad day!

NM

~Dan~ · 29-11-2004, 18:33

I'm in Leeds and I've been having trouble browsing (almost impossible) at peak times (around 5pm - 11pm) around 6 days out of 8.
I sorted it now though, just use a different webcache!
Last night someone else told me they were having problems, so I talked them through adding a proxy webcache and that cleared things up.
It seems that when I turn off the proxy webcache, the default webcache I get is ALWAYS the same one. If users were evenly distrubuted across the webcaches to spread the load then these problems wouldn't happen.

rdhw · 29-11-2004, 18:55

Quote:

Originally Posted by ~Dan~

It seems that when I turn off the proxy webcache, the default webcache I get is ALWAYS the same one. If users were evenly distrubuted across the webcaches to spread the load then these problems wouldn't happen.

See my reply to your same point in another thread. The load-balancing within a cluster is done by destination URL, not by user. If you did the load-balancing by user, and there were ten members in the proxy cluster, then the cluster would need 10 cached copies of each popular site. Doing the load-balancing by URL, each cluster needs only one copy of each URL, and each user gets routed to the right cache for the URL requested.