Ok, my post earlier this week was slightly confusing as I mentioned one
Alpha 2000/300 with 2 SCSI buses (no probls) and the second Alpha with
a single Adaptec 1742A card - major problems. A couple responses made
me see I hadn't been clear that the flakey machine only had one bus -
nothing in binary.errlog 'cuz the drives are *gone* when the bus goes.
We *do* have a second card and have swapped it for the one we suspect
may be bad (after 4 crashes in 48 hours) but I'm now thinking we might
have made bozo errors (which also don't make it into binary.errlog :^)
[a] we're running both internal and external devices, but the board we
removed (and thought might be bad) still had it's terminator pack.
[b] if the Adaptec's bus length is 3 meters and the card uses one of
them, we're probably exceeding this in 3 internals and one of the
StorageWorks towers + a 2 foot cable (and 3 drives in the tower).
We've had difficulty getting both cards in the machine in the past, but
will probably re-attempt this again shortly to separate the internal &
external devices and reduce the effective bus length.
Thanks to:
alan_at_nabeth.cxo.dec.com
SXKAC_at_orca.alaska.edu "Kurt Carlson"
zander_at_oa-bb19-15.t.sheridanc.on.ca "Mark Zander"
root_at_s1.GANet.NET "Brad Block"
flack_at_rtp4me.ENET.dec.com "Peter Flack"
The original message included ...
> Occasionally, we're seeing CAM_ERRORs and something to the effect of
> "resetting the SCSI bus" - but nothing is making it into the log files.
> The machine completely hangs up until someone resets it - bad, as this
> is our campus news server, WWW server and the box I usually use for X.
> My initial thoughts were - "How can it log anything, the SCSI's toast."
> but I'm seeing things in scu's man page that indicate logging is not as
> impossible as I'd thought ... maybe we don't have to put up with it.
andrew. (brennan_at_hal.hahnemann.edu)
Received on Fri Apr 19 1996 - 19:48:53 NZST