What happens if you turn them both into endnotes, or both level 1 routers?
What version of winpcap do you run?
Verzonden vanaf mijn BlackBerry 10-smartphone.
Origineel bericht
Van: Jean-Yves Bernier
Verzonden: zondag 25 mei 2014 17:13
Aan: hecnet at Update.UU.SE
Beantwoorden: hecnet at Update.UU.SE
Onderwerp: Re: [HECnet] Same MAC address on different nodes
At 4:39 PM +0200 25/5/14, Hans Vlems wrote:
Would you post the output of :
Of course.
---
Node characteristics as of 30-MAR-82 00:23:45
Executor node = 10.1 (SHARK)
Identification = DECNET AT PESCADOO, Management version = 4.2.0
Host = 10.1 (SHARK), Loop count = 1
Loop length = 40, Loop with = Mixed
Incoming timer = 15, Outgoing timer = 30
NSP version = 4.0.0
Maximum links = 10, Delay factor = 32
Delay weight = 2
Inactivity timer = 30, Retransmit factor = 5
Routing version = 2.0.0, Type = Area
Routing timer = 300
Broadcast routing timer = 40, Maximum address = 1023
Maximum circuits = 1
Maximum cost = 1022
Maximum hops = 10, Maximum visits = 20
Maximum area = 63, Maximum broadcast endnodes = 32
Maximum broadcast routers = 20
Area maximum cost = 1022, Area maximum hops = 10
Incoming proxy = Disabled
Outgoing proxy = Disabled
Segment buffer size = 576
---
Node characteristics as of 30-MAR-82 00:23:31
Executor node = 10.2 (SNAKE)
Identification = DECNET AT PESCADOO, Management version = 4.2.0
Host = 10.2 (SNAKE), Loop count = 1
Loop length = 40, Loop with = Mixed
Incoming timer = 15, Outgoing timer = 30
NSP version = 4.0.0
Maximum links = 10, Delay factor = 32
Delay weight = 2
Inactivity timer = 30, Retransmit factor = 5
Routing version = 2.0.0, Type = Area
Routing timer = 300
Broadcast routing timer = 40, Maximum address = 1023
Maximum circuits = 1
Maximum cost = 1022
Maximum hops = 10, Maximum visits = 20
Maximum area = 63, Maximum broadcast endnodes = 32
Maximum broadcast routers = 20
Area maximum cost = 1022, Area maximum hops = 10
Incoming proxy = Disabled
Outgoing proxy = Disabled
Segment buffer size = 576
--
Jean-Yves Bernier
At 4:59 PM +0200 25/5/14, Johnny Billquist wrote:
I assume they all "work".
However, the speed to complete the commands might differ a lot.
You are right. They all work "past a certain time".
--
Jean-Yves Bernier
On 2014-05-25 16:41, Jean-Yves Bernier wrote:
At 4:05 PM +0200 25/5/14, Johnny Billquist wrote:
Besides playing around, which I admin is awfully fun, the reason for
using DECnet was to transfer big files (tens of megabytes). So I am out
of luck for now, but I haven't said my last word.
Like I said before. I would start by examining various counters on the
machines to get more understanding of what the problem is, and then
try to solve it from there.
I know that different network interface speeds mess things up, for
example, and I have code in my bridge to help get around the problem
if the bridge is sitting between the systems.
I suspect a timing problem.
Yes. Me too. That is why I would check various counters, to identify in more detail where the timing issues are.
How do you explain that:
NFT TI:=REMOTE::FILE
works 9 times over 10, but
NFT FOO=REMOTE::FILE
hangs every time?
The later is faster.
I do not exclude a vector mis-configuration, although the system should
be unusable.
I assume they all "work". However, the speed to complete the commands might differ a lot.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
On Sun, 25 May 2014, Johnny Billquist wrote:
Oh. That is *old*. No surprise if you got some version mismatch error then.
That's what I figured.
I've already used BRU to copy the baseline with a newer DECnet kit on it over to my ra81 image (with proper MAX and HEADERS set!) and I'm nearing completion of the SYSGEN.
Nope. Used a DECserver 200/MC with a 320 hanging off it. Logged in
fine, got directory listing, created a file, logged out...then it crashed.
Probably version related as well.
Yeah.
Oh well. I thought there was some newer DECnet version available on bitsavers. Maybe not, then...
Possible, No idea. If not...it's no big deal. See above.
Johnny
--
Cory Smelosky
http://gewt.net Personal stuff
http://gimme-sympathy.org Projects
On 2014-05-25 16:35, Cory Smelosky wrote:
On Sun, 25 May 2014, Johnny Billquist wrote:
Could be...
Did you do a netgen after building RSX?
Yes.
I used the NETKIT/DECKIT version 3
(http://bitsavers.trailing-edge.com/bits/DEC/pdp11/magtapes/rsx11mplus/BB-M4…
and
http://bitsavers.trailing-edge.com/bits/DEC/pdp11/magtapes/rsx11mplus/BB-AS…
on 4.6)
Oh. That is *old*. No surprise if you got some version mismatch error then.
Also, I hope you are not using the lat programs that someone wrote
under Unix, since they are breaking the protocols in some way. I have
never bothered to find out exactly how, but they do not play nicely
with RSX. A real DECserver works fine, though.
Nope. Used a DECserver 200/MC with a 320 hanging off it. Logged in
fine, got directory listing, created a file, logged out...then it crashed.
Probably version related as well.
Oh well. I thought there was some newer DECnet version available on bitsavers. Maybe not, then...
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
At 4:05 PM +0200 25/5/14, Johnny Billquist wrote:
Besides playing around, which I admin is awfully fun, the reason for
using DECnet was to transfer big files (tens of megabytes). So I am out
of luck for now, but I haven't said my last word.
Like I said before. I would start by examining various counters on the machines to get more understanding of what the problem is, and then try to solve it from there.
I know that different network interface speeds mess things up, for example, and I have code in my bridge to help get around the problem if the bridge is sitting between the systems.
I suspect a timing problem.
How do you explain that:
NFT TI:=REMOTE::FILE
works 9 times over 10, but
NFT FOO=REMOTE::FILE
hangs every time?
The later is faster.
I do not exclude a vector mis-configuration, although the system should be unusable.
--
Jean-Yves Bernier
Would you post the output of :
Mc ncp sho exec char
On both nodes?
( or did I miss that post)
Verzonden vanaf mijn BlackBerry 10-smartphone.
Origineel bericht
Van: Jean-Yves Bernier
Verzonden: zondag 25 mei 2014 15:54
Aan: hecnet at Update.UU.SE
Beantwoorden: hecnet at Update.UU.SE
Onderwerp: Re: [HECnet] Same MAC address on different nodes
At 9:26 AM +0200 25/5/14, Johnny Billquist wrote:
You mean "the Hardware address is whatever you set in simh before
starting". CEX will set it to "Physical address" (in DECNET parlance)
which is AA-00-04-00-(1024 x area + node).
Uh... Not really. simh do not actually change the MAC address of the
physical interface.
Of course, simh don't do DECnet. Communication EXecutive does.
The simh interface starts out with 08-00-2B-AA-BB-CC, but that is
changed to AA-00-04-00-01-28 by DECnet when it starts up, at which
point it no longer listens to 08-00-2B-AA-BB-CC. That address is
just kept around as the original hardware address of the simulated
network interface inside simh.
So we have:
Executor node = 10.1 (SHARK)
Physical address = AA-00-04-00-01-28
Line = QNA-0
Hardware address = 08-00-2B-AA-BB-CC
Executor node = 10.2 (SNAKE)
Physical address = AA-00-04-00-02-28
Line = QNA-0
Hardware address = 08-00-2B-AA-BB-CC
We still have the question "when two nodes runs on the same host, is
SET XQ MAC" necessary? In other words, does the "Hardware address"
play any role in DECnet, besides just being kept around?
Since it would be pointless to put the same address, i tried
SET XQ MAC==00:00:01:00:00:01/02, then
Executor node = 10.1 (SHARK)
Physical address = AA-00-04-00-01-28
Line = QNA-0
Hardware address = 00-00-01-00-00-01
Executor node = 10.2 (SNAKE)
Physical address = AA-00-04-00-02-28
Line = QNA-0
Hardware address = 00-00-01-00-00-02
And I see no difference.
So, a NIC may have different MAC addresses depending of the network
stack running. Am I right here?
Sortof. Your machine normally do only have one MAC address, but if
you have DECnet running, it will need to change your MAC address.
All other network protocols on your machine will then also use this
"new" MAC address.
If we would run on bare metal. They don't because simh shields it.
Maybe the "Physical address" is kept just for restoring it at
shutdown?
Well, my system is working 99%, that's not so bad :)
It is working 100%, you just have performance issues...
Besides playing around, which I admin is awfully fun, the reason for
using DECnet was to transfer big files (tens of megabytes). So I am
out of luck for now, but I haven't said my last word.
--
Jean-Yves Bernier
On Sun, 25 May 2014, Johnny Billquist wrote:
As soon as I have a little time, I'll put them all up on MIM::, along with other useful RSX stuff.
Okay!
MOIRA::CSMELOSKY$ set host mandy
%SYSTEM-F-PROTOCOL, network protocol error
MANDY:: being the PDP-11? What is MOIRA::?
Yeah. MOIRA:: is my VAXstation 4000/60.
Hmm. What version of RSX and DECnet did you install?
See other email.
Johnny
--
Cory Smelosky
http://gewt.net Personal stuff
http://gimme-sympathy.org Projects
On Sun, 25 May 2014, Johnny Billquist wrote:
Could be...
Did you do a netgen after building RSX?
Yes.
I used the NETKIT/DECKIT version 3 (http://bitsavers.trailing-edge.com/bits/DEC/pdp11/magtapes/rsx11mplus/BB-M4… and http://bitsavers.trailing-edge.com/bits/DEC/pdp11/magtapes/rsx11mplus/BB-AS… on 4.6)
Also, I hope you are not using the lat programs that someone wrote under Unix, since they are breaking the protocols in some way. I have never bothered to find out exactly how, but they do not play nicely with RSX. A real DECserver works fine, though.
Nope. Used a DECserver 200/MC with a 320 hanging off it. Logged in fine, got directory listing, created a file, logged out...then it crashed.
Johnny
--
Cory Smelosky
http://gewt.net Personal stuff
http://gimme-sympathy.org Projects
