9fans - fans of the OS Plan 9 from Bell Labs
 help / color / mirror / Atom feed
* [9fans] ip bug?
@ 2008-01-14 21:31 erik quanstrom
  2008-01-15  0:02 ` Russ Cox
  0 siblings, 1 reply; 3+ messages in thread
From: erik quanstrom @ 2008-01-14 21:31 UTC (permalink / raw)
  To: 9fans

i have this network configuration.  one network.
two sets of ip addresses.

	; cat /net.alt/iproute
	0.0.0.0         /96  65.14.39.129    4    none   0
	65.0.0.0        /128 65.0.0.0        4b   ifc    -
	65.14.39.128    /123 65.14.39.128    4i   ifc    0
	65.14.39.128    /128 65.14.39.128    4b   ifc    -
	65.14.39.131    /128 65.14.39.131    4u   ifc    0
	65.14.39.159    /128 65.14.39.159    4b   ifc    -
	65.255.255.255  /128 65.255.255.255  4b   ifc    -
	12.0.0.0        /128 12.0.0.0        4b   ifc    -
	12.51.113.0     /123 12.51.113.0     4i   ifc    0
	12.51.113.0     /128 12.51.113.0     4b   ifc    -
	12.51.113.5     /128 12.51.113.5     4u   ifc    0
	12.51.113.31    /128 12.51.113.31    4b   ifc    -
	12.255.255.255  /128 12.255.255.255  4b   ifc    -
	255.255.255.255 /128 255.255.255.255 4b   ifc    -

this is my initial arp cache

	; cat /net.alt/arp
	ether  OK       65.14.39.129                             000ff72da214                    
	ether  OK       65.14.39.130                             003018ab9f39                    

now if i

	; ip/ping /net.alt/icmp!12.51.113.7!0

i usually get one packet back and snoopy reports

	236803 ms 
		ether(s=00304881bd63 d=0030488ef4fb pr=0800 ln=78)
		ip(s=12.51.113.5 d=12.51.113.7 id=474f frag=0000 ttl=255 pr=1 ln=64)
		icmp(t=EchoReq c=0 ck=3fe2 id=6608 seq=5f12)
		dump(00000000202122232425262728292a2b2c2d2e2f303132333435363738393a3b)
	236902 ms 
		ether(s=0030488ef4fb d=00304881bd63 pr=0800 ln=82)
		ip(s=12.51.113.7 d=12.51.113.5 id=be8b frag=0000 ttl=255 pr=1 ln=64)
		icmp(t=EchoRep c=0 ck=47e2 id=6608 seq=5f12)
		dump(00000000202122232425262728292a2b2c2d2e2f303132333435363738393a3b)

all subsequent pings are unanswered.  then my arp cache looks like this

	; cat /net.alt/arp
	ether  OK       65.14.39.129                             000ff72da214                    
	ether  OK       65.14.39.130                             003018ab9f39                    
	ether  OK       12.51.113.7                              000ff72da214                    

for some reason, 12.51.113.7 is assigned the same ea as the gw for the 65. network

so the question is, am i doing something wrong or is this a mistake in the
arp code?

- erik


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [9fans] ip bug?
  2008-01-14 21:31 [9fans] ip bug? erik quanstrom
@ 2008-01-15  0:02 ` Russ Cox
  2008-01-15  0:50   ` erik quanstrom
  0 siblings, 1 reply; 3+ messages in thread
From: Russ Cox @ 2008-01-15  0:02 UTC (permalink / raw)
  To: 9fans

> now if i
> 
> 	; ip/ping /net.alt/icmp!12.51.113.7!0
> 
> i usually get one packet back and snoopy reports

there are surely more packets than that.
for example, where are the arp packets that initially
discover that 12.51.113.7 is 0030488ef4fb?

i suspect that the machine is sending out an arp request
for 12.51.113.7 and getting back responses both from
the actual 12.51.113.7 as well as the 65 gateway.  
the 65 gateway responds a little bit later, so that you 
manage to get one ping off between getting the first
response and the second response.  if this is the case,
then your 65 gateway is wrong.

alternately, it could be that the 65 gateway is somehow
getting the ping traffic and responding with a 
"no route to host" and somehow that response is
poisoning the arp cache.  

snoopy -f 'icmp || arp' would tell for sure.
it would also be nice to see the output of netstat -i.

russ


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [9fans] ip bug?
  2008-01-15  0:02 ` Russ Cox
@ 2008-01-15  0:50   ` erik quanstrom
  0 siblings, 0 replies; 3+ messages in thread
From: erik quanstrom @ 2008-01-15  0:50 UTC (permalink / raw)
  To: 9fans

On Mon Jan 14 19:03:58 EST 2008, rsc@swtch.com wrote:
> > now if i
> > 
> > 	; ip/ping /net.alt/icmp!12.51.113.7!0
> > 
> > i usually get one packet back and snoopy reports
> 
> there are surely more packets than that.
> for example, where are the arp packets that initially
> discover that 12.51.113.7 is 0030488ef4fb?
> 
> i suspect that the machine is sending out an arp request
> for 12.51.113.7 and getting back responses both from
> the actual 12.51.113.7 as well as the 65 gateway.  
> the 65 gateway responds a little bit later, so that you 
> manage to get one ping off between getting the first
> response and the second response.  if this is the case,
> then your 65 gateway is wrong.

your diagnosis is excellent as usual.  this is exactly the case.
i have no access yet to the 65 gateway so it's hard telling
what it's problem is.  it does not do the same thing to
other 12 addresses.

- erik


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2008-01-15  0:50 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2008-01-14 21:31 [9fans] ip bug? erik quanstrom
2008-01-15  0:02 ` Russ Cox
2008-01-15  0:50   ` erik quanstrom

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).