[Info-vax] hub better than switch?!

[Jan-Erik Soderholm]

Phillip Helbig---undress to reply wrote 2012-03-04 10:31:
> In article<jiv9fo$k7f$1 at news.albasani.net>, Jan-Erik Soderholm
> <jan-erik.soderholm at telia.com>  writes:
>
>>   >  about a slowly (about 2 per day) increasing error count,
>>
>> Does it say anything about *what* error ?
>
> PEA0:                                   46
>

OK, right. :-)  I thought it whould be possinly to get some
specific *error*. You have mentioned the *device* several
times now. :-)

It might be interesting to see if it is something like
"...lost..." or "...overrun..." or completely different.


> This is the one which is increasing, but only on the node with the 100
> Mb/s network card.
>
>> If I understand correctly, you have VMS-Cluster ovar LAN with one node
>> running at 100 Mb/s and the other node/nodes running at 10 Mb/s ?
>
> Right.
>
>> What if you, just as a test, force the 100 Mb/s node down to 10 Mb/s ?
>
> I suppose I could, but if it works, it's not something I want.  I would
> like each machine to run at the speed it is capable of and had hoped
> that such a switch would allow this.
>
>>   >  while on the "hub" it ran at 10...
>>
>> So it that case *all* cluster members run at the same speed ?
>
> Right, though other things (a WLAN access point and another switch) ran
> at 100.
>
>> I'd look closer to the speeds then to any hub/switch diffrences.
>
> The thing is that with the "hub" (not really a hub---see previous posts)
> the problem stopped.  OK, maybe the problem is due to the mixed speeds
> and the "hub" forced the one node to slow down to 10 Mb/s.  However,
> normally it shouldn't be a problem to have different ethernet speeds in
> a LAN cluster, right?
>

I have absolutely no idea! :-)

If you run some large FTP transfer between these nodes, does that run
fine without increasing the error count ?

Jan-Erik.