Conference netcad::hub_mgnt

783.0. "DECrepeater 90C Timing problems ?" by FRSIT::MAYER () Wed Mar 02 1994 12:27

Hi, has anybody an idea about this.

"Also crossposted to ethernet notes conference"
    
I saw simialar problems as described in Ethernet Note 1085, 231 and HUB_Mgt
Note 14, that DR90C and DR90T is less tolerant for out-of-spec timing as old
repeaters like DEMPR and DESPR.

But we have an additional symptom, where we havn't any explanation. May somebody
has an idea whats going on.

Here are my configuration:

In the Computer Room there are several VAX's (66xx, 4xxx, 8250) and also two
Alpha's connected via an DELNI to thickwire backbone. Additional there are two
VAX8530 connected to an other DELNI, which is connected to the thickwire
backbone via an LANbridge 150.

In the floors the customer has moved from DEMPR/DESPR configuration to DEChub 90
with DECrepeater 90T. On the Thinwire segments the customer is using PC's and
MACintosh's (DECnet). Befor changing from DEMPR to DEChub's there are no
problems at all. In the moment he is using the DEChub's he got problems with
the MACintosh's during filetransfer, if the destination is one of the Systems
connected to the backbone via delni, he has no problems if the destination is
one of the VAX8530's.

I did an Line trace direct on the DR90T thinwire segment where the MAC is
connected. There are no corrupted packets detected, but in the moment the MAC
is using large packets, there are no ack's from the destined VAX's.

I did a second line trace by connecting the HP to an different repeater port, 
and in moment the MAC is transmitting large frames, the HP detects them as
missaligned (5555 pattern starting at about byte 1200). This occurs only if
the destination is one of the VAX's which are connected to the backbone via
the DELNI. No problem if the destination is one of the VAX8530.

The same results if I connect the HP between the DEChub and the Thickwire
Backbone.

If I connect the MAC to the DEChub via an DESPR everything works fine.

If we changed the Ethernet Card's in the MAC's to an different type also,
everything works fine. So I believ we have a serious clocking problem.

BUT WHY DOES THE REPEATED PACKETS GOT CORRUPTED IF DESTINIED TO SOME NODES AND
NOT CORUPPTED TO OTHER NODES, EVEN IF THEY ARE USING THE SAME PATH VIA THE
REPEATER PORTS AND THICKWIRE CABLE ??


regards Juergen