> >> The machine has 2 interfaces (obviously ep0 and ep1), and
> >> after about 24 hours uptime, ep1 was somehow broken:
> >>
> >> If trying to ping: sendto: no bufferspace available
> >>
> >> ifconfig ep1 down, ifconfig ep1 up was a work-around, but no
> >> solution...
> >>
> >> Netstat showed nothing 'extraordinary' like high ammount of
> >> collissions, late collisions, fragmentations, carrier losses
> >> even the net work load was almost a little above nothing.
>
> >All of these are typical of media problems - bad cables or
> >ports on your hub or switch.
> I can exclude this:
> - The statistics do not show any 'loss of carrier' (mentioned
> above)
> - On the same hub's, there are some highly available financial
> applications and web-servers. If there would be that sort of
> problem, I'd surely noticed.
> - For some other reason, we changed the hub and cabeling, and
> the problem still exists...
Ack! My error - I didn't notice the "nothing".
> >> I remember to have the same problem with another box running
> 3.3
> >> generic (other hardware...) - so I am wondering if the driver
> >> might be somehow broken.
> >
> >But if it was a driver problem, the problem would show up on
> >both interfaces, right?
>
> Yes, and inbetween I can confirm: The problem occurs on both
> interfaces; the chance that the problem occurs is slightly
> higher on the interface with the higher network load.
>
> To sum up: As the very same problem occurs with pretty different
> hardware, I really think there's something wrong with the
> cooperation of the ep-driver and 3com 509 - maybe the driver,
> maybe some design error on the network cards...
>
> So if sb says: 'No probs with ep driver and ne2000 cards', then
> I think we have to ask 3com... ;)
Aren't the 509s ISA-based? My thought would be just to replace them with a
newer, more stable NIC.
--
Matt Emmerton
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-net" in the body of the message
