Re: DECnet phase IV node drops itself. Why? How?

From: AEF (spamsink2001_at_yahoo.com)
Date: 10/05/05 

Date: Wed, 05 Oct 2005 01:39:19 GMT

Hoff Hoffman wrote:
> In article <1128004369.768012.272660@g44g2000cwa.googlegroups.com>, "AEF" <spamsink2001@yahoo.com> writes:
> :Hoff Hoffman wrote:
> :> In article <1127938795.354636.197410@g43g2000cwa.googlegroups.com>, "AEF" <spamsink2001@yahoo.com> writes:
> ..
> :> What is connected onto the SVA-0 port here?
> :
> :Other MicroVAX 3100's, mostly model 80's, but 3 model 95's.
>
> You're using the integrated NIC on the MicroVAX 3100, then.
> (There are various pieces of hardware that use the DECnet
> SVA interface.)
>
> As for the question, I was refering to the controller hardware
> and the hardware immediately hanging off the controller, as well
> as any bridges, switches, hubs, repeaters or terminators that
> might be in use here.

We have a 10 Base - T transceiver (IEEE 802.3) attached to the network
port on the back of the MicroVAX. Some of them have redundancy. That's
here in NYC. I assume it's pretty much the same in London (I've never
been there and no one is there right now to help me answer any such
questions.)

The cable in the transceiver goes to a Cisco switch. This is true for
both NYC and London systems.

I hope this answers your questions. Sorry, but this is my weak area.

> :> Is this box an end-node, a level one router, or an area router?
> :
> :NODE_X is a router
>
> Is there more than one circuit connection on these hosts?
> (On a MicroVAX 3100 series box, that would be comparatively
> unusual -- but there are ways to achieve it.)

Multiple physical circuits? Only with the redundancy transceivers in
which case there are there two cables going to separate switches.

$ NCP SHOW KNOWN CIRC

Known Circuit Volatile Summary as of 29-SEP-2005 18:44:41

   Circuit State Loopback Adjacent
                                              Name Routing Node

  DNIP-0-0 on 1.aaa (XXXX5)
  SVA-0 on 1.bbb (NODE_F)
  SVA-0 1.ccc (xxxxx7)
  SVA-0 1.ddd (xxxxx8)

Node 1.aaa is in NYC. The others are all in London on the same VLAN.

> :These are on a VLAN in London.
>
> So there is some comparatively unusual hardware within the network
> here -- these systems are not locally connected.

They are probably all connected to the same pair of Cisco switches
(assuming they have the redundancy transceivers). I login in to them
via telnet from NYC. All of the affected nodes are in London.

> :We have 2 DNIP tunnels, one from NYC to London and another from NYC to
> :Hong Kong. NODE_X is one end of the NYC to London tunnel.
>
> So there could easily be a problem within the connection between
> the hosts.

The DNIP tunnels are needed *only* for DECnet connectivity from one
city to another (overseas, actually -- the cities are NYC, London, and
Hong Kong). All the affected systems are in London and in no way use
DNIP to send and receive DECnet packets from each other.

>
> Are the VLAN devices (bridges or routers, I assume) reporting any
> connectivity errors?

I'm waiting to hear back from the network group.

But suppose there were connectivity errors. Could that have caused a
node to drop itself? What does it mean for a node to drop itself? And
why would a node reject another (garbled password?)?

>
> :> Does this box have more than one network connection to the
> :> same or to the same bridged network segment?
>
> I'll repeat: does this box have more than one network connection
> to the same or to the same bridged network segment? DECnet
> Phase IV does not like that, and will generate errors -- usually
> IVADDR invalid media address errors IIRC, but things can and do
> get strange when the node can "see" itself.

I don't think so. The only case of more than one is the redundancy
transceivers. NOTE: This system works fine the vast majority of the
time. I was mostly interested in what it means for a node to drop
itself. How can a node drop itself? And the same for rejection.

>
> :> What versions of DECnet and OpenVMS (and ECOs) are involved here?
> :
> :NODE_F: VMS 6.1
> :NODE_X: VMS 6.2
> :NODE_Y: VMS 6.1
>
> Old, obviously. No hope of upgrades or ECOs, I assume.

Well, yes and no. I am planning to upgrade the 6.1 systems to 6.2, but
I want to assemble a fully patched system disk and test our app on it.
The existing 6.2 systems are running with zero or very few patches.
When they were set up I was told to not put patches on them. The
problem with upgrades is that I have 34 systems on line. That's a lot
of work! And most of them are overseas.

>
> ::> Network cabling problems (cable shorts, too little or too much
> ::> termination on ThinWire or thick wire, etc) and controller (and
> ::> the transceiver, for those NICs that use one) hardware problems
> ::> can cause DECnet circuits to yo-yo.
> ::
> ::
> ::Well, most of the time things are fine. I was curious, however, how a
> ::node can drop itself?! There may be some off-hours work being done in
> ::our London data center (I'm located in NYC) as they are preparing for
> ::relocation to another site in London.
>
> When the network connection becomes unstable due to faults within
> the cabling, the VLAN (in this case), or a hardware problem, the
> host can and will report errors.
>
> Messing about within a network can most certainly trigger these
> errors -- this is why I have been asking about connectivity here.

So it could be a bad transceiver?

Thanks again for your help and your patience.

Bonus question:

I was asked if I could have 2 NIC cards on the MicroVAX. This would be
in case we decide to chuck the tape drives (well, save them for
restores) and do backups via NetBackup over the network. They want a
primary network connected and the backup network connected to each
MicroVAX that needs to be backed up. But we might try it even if it's
not possible, as the backups are relatively small (compared to our
other non-VAX systems).

Thanks again.

Relevant Pages

  • Re: DECnet phase IV node drops itself. Why? How?
    ... > (There are various pieces of hardware that use the DECnet ... > So there could easily be a problem within the connection between ... The DNIP tunnels are needed *only* for DECnet connectivity from one ... I'm waiting to hear back from the network group. ...
    (comp.os.vms)
  • Re: windows/internet explorer suddenly closes
    ... > 10/100M cable communicating by a network card. ... > actually, i remember even there's no network connection, ... > the file explorer also can close itself. ... that indicates a possible hardware problem. ...
    (microsoft.public.windowsxp.general)
  • Re: Making PPP device persist, even after lost connection?
    ... of hardware and a device driver for said hardware. ... virtual network connection over some sort of physical transport. ... the serial port doesn't disappear, ... but a serial port isn't a network device. ...
    (comp.os.linux.networking)
  • Re: NIC Having Multiple IP Addresses?
    ... As long as you're network cable is connected ... have a HARDWARE connection. ... And mii-tool will find an active HARDWARE connection. ...
    (comp.os.linux.networking)
  • Re: DECnet phase IV node drops itself. Why? How?
    ... I was refering to the controller hardware ... :These are on a VLAN in London. ... So there is some comparatively unusual hardware within the network ... So there could easily be a problem within the connection between ...
    (comp.os.vms)