MTU Fragmentation issue showing in AIX traceroute

From: steven_nospam at Yahoo! Canada (steven_nospam_at_yahoo.ca)
Date: 03/15/05 

Date: 15 Mar 2005 06:09:12 -0800

Hi All,

I have a problem at a customer's remote site. I know my AIX but am a
bit rusty on the networking end.

What was reported is that the network printers at the site are really
slow, show SENDING in the queues for a long time, and occasionally go
into a DOWN state.

A check of the settings and the network finally turned up the following
warnings in the traceroute command:

# traceroute 192.168.80.25
trying to get source for 192.168.80.25
source should be 192.168.77.26
traceroute to 192.168.80.25 (192.168.80.25) from 192.168.77.26
(192.168.77.26),
30 hops max
outgoing MTU = 1500
 1 192.168.77.3 (192.168.77.3) 4 ms 7 ms 3 ms
 2 25.176.38.58 (25.177.38.58) 5 ms
fragmentation required, trying new MTU = 1492
 2 * 5 ms
fragmentation required, trying new MTU = 1480
 2 * 5 ms
fragmentation required, trying new MTU = 1472
 2 * 113 ms
fragmentation required, trying new MTU = 1006
 2 utahlp4 (192.168.80.25) 149 ms 89 ms 96 ms

I understand that AIX is performing PMTU discovery and altering the MTU
size down until it hits on a value that is acceptable to both the
sender and the receiver.

What seems to be doing this is a Cisco router or some equipment at the
remote site. However, I am not getting too much support for this theory
from the network admin at the remote site.

I was going to contact our IBM AIX Support desk until I noted that the
fragmentation is not coming from OTHER remote branches when I
traceroute from the RS/6000. I may do so anyway for a second opinion.
Just wondering if anyone has ever encountered this and has some
thoughts on what I can have them check?

TIA for any advice or suggestions.

Steve

Relevant Pages

  • Re: AD Sites and Services Question
    ... Tested the set logonserver and it is showing the DC on the remote site. ... I had one old NS record that I deleted for a server that no longer existed. ... site was created in case we ever did stick an AD server into that network. ... Run dcdiag, netdiag and repadmin in verbose mode. ...
    (microsoft.public.windows.server.active_directory)
  • Re: AD Security Groups break Authentication
    ... account only in Domain Users, indicates something else is surely going ... login using my own account & a couple others, ... We tried reducing the MTU on our routers this morning to 1400. ... Microsoft network server: Digitally sign communications ...
    (microsoft.public.windows.server.active_directory)
  • Re: AD Security Groups break Authentication
    ... Changing the MTU on the network equipment was a bad idea. ... It would take 15+ minutes to login and there was a 1-2 minute delay each time we tried to access resources. ... account only in Domain Users, indicates something else is surely going ...
    (microsoft.public.windows.server.active_directory)
  • Re: Remote site bandwidth problems
    ... > 1MB ADSL link at the remote site. ... but the problem is that the network performance is being ... > ADSL link is constantly lit, so it does look like a network bottleneck. ... All of the remote site client computers are members of the SBS ...
    (microsoft.public.windows.server.sbs)
  • RE: TCP/IP Stack Hardening
    ... When PMTU is disabled, the default MTU is 1500 for local subnet and 576 for ... disabling PMTU is probably not the most important network stack ... I've not found disabling ICMP redirects a problem as long as the routing ...
    (Focus-Microsoft)