| Title: | DEChub/HUBwatch/PROBEwatch CONFERENCE |
| Notice: | Firmware -2, Doc -3, Power -4, HW kits -5, firm load -6&7 |
| Moderator: | NETCAD::COLELLA DT |
| Created: | Wed Nov 13 1991 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 4455 |
| Total number of notes: | 16761 |
According to the Redundant Fiber-Optic Link Specifications, the Complex
Redundant link configuration could be as follows:
DECRepeater 900FP DECRepeater 90FS --
|
Master Port ------------------------ Responder Port |
|Thinwire
|(Multistack)
|
DECRepeater 90FS --
Master Port ------------------------ Responder Port
And it is required that the master ports must reside on the same module.
If we forget the redundancy specifications and we make the connection
using 4 different modules, obviously we have an ethernet loop and as
consequence of the collisions one segment is automatically
Auto-partitioned. It this situation, one segment is working as "active"
and the other is OFF because it has been autopartitioned. If the
"active link" is broken the other link becomes active automatically.
In that case we should be able to have a "more complex (total)"
redundant configuration. (My customer has tested this configuration and
it works)
- What is wrong?, Why it is not recommended? Why redundancy against
automatic autopartitioning?. As I don't know how auto-partitioning
works, I cannot answer all these questions?
- By the way, Is it needed to assign an IP address to both DECrepeaters
modules in order to configure the ports as responders (using HUBwatch)
although both modules are in the same Multistack? Note: 2085 why?
Thanks in advance for your help,
Jose Luis Platas
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 3429.1 | Can't guarantee that it will work | NETCAD::PAGLIARO | Rich Pagliaro, Networks BU, HPN | Mon Apr 08 1996 09:57 | 33 |
There are a number of potential problems in using auto-partitioning to
provide fault tolerance. The main issue is that the resulting
configuration can be unstable and is not deterministic. That is, you
can't predict which ports (or how many ports) will auto-partition.
Consider your example of using four modules connected in a loop and
with redundancy not enabled.
DECRepeater 900FP #1 DECRepeater 90FS #1 --
|
Port 1 ----------------------------- Port 1 |
Port 2 Link 1 Port 2 |Thinwire
: Port 3 |(Multistack)
+---- Port 12 |
| DECRepeater 90FS #2 -- Link 3
| DECRepeater 900FP #2
| Port 1 ----------------------------- Port 1
| Port 2 Link 2 Port 2
| : Port 3
+---- Port 12
If the above 90FS Thinwire ports auto-partition, this effectively makes
the MultiStack thinwire segment unusable. Therefore, all of the modules
in the stack will lose network connectivity with one another. This
would be true if the 90FS's were configured in a DEChub 90 or DEChub
900.
-Rich
| |||||
| 3429.2 | Thanks and two IP address? | IB002::PLATAS | Mon Apr 08 1996 11:32 | 14 | |
Rich,
Thanks a lot for your answer, I agree with you, this configuration is
not stable and we cannot predict which ports and how many ports!!! will
be partitioned.
Regarding the second question is it required to assign two IP address
for both repeaters althoght they are on the same Multistack in order to
configure the responders ports? Why?
Thanks again for your help.
| |||||
| 3429.3 | NETCAD::DRAGON | Tue Apr 16 1996 09:21 | 13 | ||
Jose,
It is required that both repeaters have IP addresses assigned to
them in order to configure the responder ports. The reason for this
is that not all MIB support is provided for the repeaters via proxy
access (stack master). This includes DPR support. Therefore, you
must configure DPR by accessing the modules own SNMP agent.
Sorry for late response. I was away.
Regards,
Bob
| |||||