olling@tcgould.tn.cornell.edu (Cliff Olling) (01/07/89)
A while back I posted a plea for advice on why our newly installed Maxtor
XT-8760E 760Mb (unformatted) disk was over eight times slower than our
existing Sun-supplied disk. This was on a 3/160C running 3.5. First let
me give my thanks to all of those who replied. Now it's time to report
the conclusions we've reached.
The slowness appears to have been due to a combination of infant mortality
(Maxtor concluded that we had received a partially dead disk) and us
strapp- ing the disk incorrectly. The disk now appears to run reasonably
well in a quite optimistic (read: "big") geometry for what we paid for it.
To wit,
/etc/dkinfo sd1g
Emulex MD21 controller at addr 200000, unit # 1
1626 cylinders 15 heads 53 sectors/track 1292670 sectors
df /usr3
Filesystem kbytes used avail capacity Mounted on
/dev/sd1g 618573 110986 507587 18% /usr3
As I said, the disk _is_ working, but about once a day we see some variant
on the following error messages, and about 50 messages when we do dumps:
sd1g: read recoverable, blk 168711 (abs blk 168711)
sense key(0x1): soft error, error code(0x16): dma timeout
sd1g: read recoverable, blk 171707 (abs blk 171707)
sense key(0x1): soft error, error code(0x17): recoverable error
sd1g: read recoverable, blk 181026 (abs blk 181026)
sense key(0x1): soft error, error code(0x16): dma timeout
sd1g: read recoverable, blk 347952 (abs blk 347952)
sense key(0x1): soft error, error code(0x17): recoverable error
Maxtor suspects that the "timeout"'s are due to having one large partition
on the disk instead of a few smaller ones. Once again, I have to ask if
anyone has any advice to offer.
Thanks in advance,
Clifford Olling, olling@tcgould.tn.cornell.edu, Cornell University
Modern Languages and Linguistics
cornell!plab!olling