🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Thread View: comp.protocols.ppp
2 messages
2 total messages Started by Ross Porter Wed, 22 Jul 1992 00:58
PPP client routing
#19
Author: Ross Porter
Date: Wed, 22 Jul 1992 00:58
91 lines
4013 bytes
We suspect that some of the problems we are having with our PPP connection:

    PPP HOST (SPARC)        PPP CLIENT (SPARC)
  130.91.68.47(turock) <--> 130.91.68.44(sum)

are due to improper routing.  One symptom of the problem is that
while rsh client->host works fine, rsh client->client yields
a ~75sec delay.  Once either connection is established, 
things work fine.  We strongly suspect that the wrong routes
are being used.  

rsh client->host behaves normally and hits on only two routes:

  sum(130.91.68.44)> NetStatDiff rsh 130.91.68.47 ls
  destination	gateway		flags	refcnt	use		interface
  127.0.0.1	127.0.0.1	UH	1 -> 2	98541 -> 98550	lo0
  130.91.68.47	130.91.68.44	UH	1 -> 1	66139 -> 66151	ppp0

rsh client->client takes ~75sec and hits on *four* routes:

  sum(130.91.68.44)> NetStatDiff rsh 130.91.68.44 ls
  destination	gateway		flags	refcnt	use		interface
  127.0.0.1	127.0.0.1	UH	1 -> 1	98550 -> 98559	lo0
  130.91.68.44	130.91.68.44	UH	0 -> 2	32255 -> 32275	le0
  130.91.68.47	130.91.68.44	UH	1 -> 1	66221 -> 66260	ppp0
  default	130.91.68.47	UG	0 -> 0	1124 -> 1136	ppp0

The 44->44 route through le0 looks strange to me since there's
no ethernet cable on le0, but Karl Fox's INSTALL/README example
has something similar.

System info, including complete route tables, for both systems
can be found below.

A similar situation, telnet client->client yields the same
delay, but the "connected to <client>" message appears in ~1sec.
This suggests that there these connections can be easily be made:

  client -> host   (e.g. send string "ls")
  client <- host   (e.g. send result of ls)
  client -> client (e.g. send string "ls")

However, this connection takes ~75sec to establish:

  client <- client (e.g. send result of ls)

The results of e.g. trace rlogin agree with this supposition.

Further symptoms (which we think are related) include:

  - openwin is dog-ass slow to load
  - name resolution for systems in /etc/hosts is slow (~75sec)

Any help would be greatly appreciated.

Ross Porter
ross@psych.upenn.edu

-----------------

The client's full route table:

  sum(130.91.68.44)> netstat -rn
  Routing tables
  Destination          Gateway              Flags    Refcnt Use        Interface
  130.91.68.32         130.91.68.47         UGH      1      29401      ppp0
  127.0.0.1            127.0.0.1            UH       1      98550      lo0
  130.91.68.44         130.91.68.44         UH       0      32255      le0
  130.91.68.31         130.91.68.47         UGH      1      53065      ppp0
  130.91.68.47         130.91.68.44         UH       1      66199      ppp0
  default              130.91.68.47         UG       0      1124       ppp0
  130.91.68.0          130.91.68.44         U        0      15501      le0

The host's full route table:

  turock(130.91.68.47)> netstat -rn
  Routing tables
  Destination          Gateway              Flags    Refcnt Use        Interface
  127.0.0.1            127.0.0.1            UH       1      4234       lo0
  130.91.68.44         130.91.68.47         UH       3      76201      ppp0
  default              130.91.68.1          UG       1      19355      le0
  130.91.68.0          130.91.68.47         U        10     518569     le0

Note that I am deviating from Karl Fox's INSTALL/README suggestions.
INSTALL/README suggests adding a static route to the gateway machine:

  gateway> route add net PPPClientSubnet PPPhost

Unfortunately, I do not have control over our subnet router.

client and host are both SPARC IPC's running pppd-1.01beta.
Re: PPP client routing
#22
Author: Ross Porter
Date: Thu, 23 Jul 1992 13:55
17 lines
1302 bytes
>We suspect that some of the problems we are having with our PPP connection
>are due to improper routing.  One symptom of the problem is that
>while rsh client->host works fine, rsh client->client yields
>a ~75sec delay.  Once either connection is established, 
>things work fine.  

A kind soul suggested changing the localhost entry in the PPP client (sum):

  original: 127.0.0.1       localhost
  new:      127.0.0.1       localhost sum.psych.upenn.edu sum.psych sum

A wonderful hack.  This does work, although it creates a few annoyances.
I'm curious as to why the original localhost entry worked for ethernet,
but didn't for PPP.  Is it because the POINTOPOINT network interface
type takes precedence over localhost/loopback?  If so, is this desirable?

Ross
Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads