7 May
2020
7 May
'20
4:59 a.m.
Does someone have a VMS system on HECnet with a guest login that I can
use? I need to log in and then SET HOST out again.
I'm working on fixing a few bugs in the DECnet/Linux NML and I need to
make a connection to my Linux machine from a node outside my local area for
testing purposes.
Thanks,
Bob
7 May
7 May
5:05 a.m.
Hi!
There's a BOPOHA host which is reachable via telnet at pdp-11.online:5023.
GUEST without password.
On 07-May-2020 09:59, Robert Armstrong wrote:
? Does someone have a VMS system on HECnet with a
guest login that I can use?? I need to log in and then SET HOST out again.
? I?m working on fixing a few bugs in the DECnet/Linux NML and I need to make a
connection to my Linux machine from a node outside my local area for testing purposes.
Thanks,
Bob
6:24 p.m.
Hi Bob,
I don?t know what fixes your adding to NML but I?ve just got ?show active circuits? and
?show active nodes? support in place. The latest code is available on github:
<https://github.com/JohnForecast/RapsbianDECnet
<https://github.com/JohnForecast/RapsbianDECnet>>
John.
On May 6, 2020, at 10:59 PM, Robert Armstrong <bob
at jfcl.com> wrote:
Does someone have a VMS system on HECnet with a guest login that I can use? I need to
log in and then SET HOST out again.
I?m working on fixing a few bugs in the DECnet/Linux NML and I need to make a
connection to my Linux machine from a node outside my local area for testing purposes.
Thanks,
Bob
6:53 p.m.
John Forecast <john at forecast.name> wrote:
I?ve just got ?show active circuits? and ?show active nodes? support ...
Sounds like we're working on the same thing. I started out making a few changes to
make Paul's HECnet mapper happy and so far I've
* Fixed some careless casting of void* to integers that were causing compiler warnings.
* Rewrite send_exec() so that SHOW EXEC and SHOW EXEC CHAR return items more in line
with other DECnet implementations.
* Fix SHOW KNOWN nodes so it sends the executor first and doesn't repeat it in the
list.
* SHOW ADJACENT nodes should not send the executor - we're not adjacent to
ourselves!
* Implement SHOW KNOWN|ACTIVE LINES|CIRCUITS
* currently working on implementing SHOW ACTIVE NODES
You can try it on ZITI:: if you want.
BTW, how did you implement ACTIVE NODES (i.e. how did you determine which nodes are
"active")?
Bob
9:08 p.m.
On May 7, 2020, at 12:53 PM, Robert Armstrong <bob
at jfcl.com> wrote:
Unfortunately I don?t have any nodes on HECnet, I should really fix that.
John Forecast <john at forecast.name>
wrote:
I?ve just got ?show active circuits? and ?show active nodes? support ...
Sounds like we're working on the same thing. I started out making a few changes to
make Paul's HECnet mapper happy and so far I've
* Fixed some careless casting of void* to integers that were causing compiler warnings.
* Rewrite send_exec() so that SHOW EXEC and SHOW EXEC CHAR return items more in line
with other DECnet implementations.
* Fix SHOW KNOWN nodes so it sends the executor first and doesn't repeat it in the
list.
* SHOW ADJACENT nodes should not send the executor - we're not adjacent to
ourselves!
* Implement SHOW KNOWN|ACTIVE LINES|CIRCUITS
* currently working on implementing SHOW ACTIVE NODES
You can try it on ZITI:: if you want.
BTW, how did you implement ACTIVE NODES (i.e. how did
you determine which nodes are "active?)?
The above list looks very similar to what I have running. ACTIVE NODES for a
nonrouting node is defined as all nodes at the other end of a logical link, adjacent nodes
and the designated router.
John.
Bob
10:14 p.m.
On May 7, 2020, at 3:08 PM, John Forecast <john at
forecast.name> wrote:
I include the executor in active nodes, and that's generally what I've seen others
do. The spec does not say so, interestingly enough. The spec is also confused because
"adjacent nodes" isn't defined on non-routing nodes; the only adjacency they
have is the designated router. I'm assuming it wasn't the intent for this to show
the "on-NI cache", I don't remember seeing anyone do that.
paul
BTW, how did you implement ACTIVE NODES (i.e. how
did you determine which nodes are "active?)?
The above list looks very similar to what I have running. ACTIVE NODES for a
nonrouting node is defined as all nodes at the other end of a logical link, adjacent nodes
and the designated router.
10:26 p.m.
Paul Koning <paulkoning at comcast.net> wrote:
I include the executor in active nodes, and that's generally what I've seen others
do.
That's what VMS does, and it follows the rule of "executor is always reported
first", and also includes additional data for the executor (identification, active
links, state).
"adjacent nodes" isn't defined on
non-routing nodes
the only adjacency they have is the designated router
Is that really true? Suppose an end node had multiple circuits - it's true that it
can't route between nodes on different circuits, but that node can still access all
those circuits, right? Or is an end node effectively limited to one circuit only?
Bob
10:35 p.m.
On 2020-05-07 22:26, Robert Armstrong wrote:
Endnodes can only use one circuit. If you have two enabled, the other
one is a hot standby in case the first one goes down.
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
Paul Koning <paulkoning at comcast.net>
wrote:
I include the executor in active nodes, and that's generally what I've seen
others do.
That's what VMS does, and it follows the rule of "executor is always reported
first", and also includes additional data for the executor (identification, active
links, state).
"adjacent nodes" isn't defined on
non-routing nodes
the only adjacency they have is the designated router
Is that really true? Suppose an end node had multiple circuits - it's true that
it can't route between nodes on different circuits, but that node can still access all
those circuits, right? Or is an end node effectively limited to one circuit only?
11:07 p.m.
On May 7, 2020, at 4:14 PM, Paul Koning <paulkoning
at comcast.net> wrote:
SHOW ACTIVE NODES on both RSX and VMS does the equivalent of SHOW EXECUTOR SUMMARY
followed by all the other ?active? nodes. I don?t think anyone used the ?on-NI cache? but
it looks as though both RSX and VMS keep a cache of recently computed round-trip delays
and use that for ?adjacent nodes?.
John.
On May 7, 2020, at 3:08 PM, John Forecast
<john at forecast.name> wrote:
I include the executor in active nodes, and that's generally what I've seen
others do. The spec does not say so, interestingly enough. The spec is also confused
because "adjacent nodes" isn't defined on non-routing nodes; the only
adjacency they have is the designated router. I'm assuming it wasn't the intent
for this to show the "on-NI cache", I don't remember seeing anyone do that.
paul
BTW, how did you implement ACTIVE NODES (i.e. how
did you determine which nodes are "active?)?
The above list looks very similar to what I have running. ACTIVE NODES for a
nonrouting node is defined as all nodes at the other end of a logical link, adjacent nodes
and the designated router.
11:38 p.m.
On 2020-05-07 23:07, John Forecast wrote:
Exactly what is included in "ACTIVE NODES" and "ADJACENT NODES" in RSX
is a topic I've discussed with Paul before.
DECnet actually do have a formal definition for these, but RSX is not
completely following it.
For ACTIVE NODES, it's adjacent nodes, as well as any node for which RSX
have some statistics (so yes, the round trip delay is a sign that it
have statistics).
ADJACENT NODES on the other hand, are nodes that you have a direct
connection to, and/or are online.
So, any nodes that are at the other end of any point to point circuit,
as well as any nodes that are online on an ethernet that the RSX machine
is also on.
So there are nodes that might not be adjacent, but might be active. But
the other way around cannot happen.
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 May 7, 2020, at 4:14 PM, Paul Koning
<paulkoning at comcast.net> wrote:
SHOW ACTIVE NODES on both RSX and VMS does the equivalent of SHOW EXECUTOR SUMMARY
followed by all the other ?active? nodes. I don?t think anyone used the ?on-NI cache? but
it looks as though both RSX and VMS keep a cache of recently computed round-trip delays
and use that for ?adjacent nodes?. On May 7, 2020, at 3:08 PM, John Forecast
<john at forecast.name> wrote:
I include the executor in active nodes, and that's generally what I've seen
others do. The spec does not say so, interestingly enough. The spec is also confused
because "adjacent nodes" isn't defined on non-routing nodes; the only
adjacency they have is the designated router. I'm assuming it wasn't the intent
for this to show the "on-NI cache", I don't remember seeing anyone do that.
paul
BTW, how did you implement ACTIVE NODES (i.e. how
did you determine which nodes are "active?)?
The above list looks very similar to what I have running. ACTIVE NODES for a
nonrouting node is defined as all nodes at the other end of a logical link, adjacent nodes
and the designated router.
1660
days inactive
1660
days old
9 comments
5 participants
participants (5)
-
bob＠jfcl.com -
bqt＠softjar.se -
form＠pdp-11.online -
john＠forecast.name -
paulkoning＠comcast.net