Date: Fri, 3 Jan 1997 12:52:42 -0600 (CST) From: peter@taronga.com (Peter da Silva) To: wilko@yedi.iaf.nl (Wilko Bulte) Cc: peter@taronga.com, scsi@freebsd.org Subject: Re: Exabyte 8200 Message-ID: <199701031852.MAA13083@bonkers.taronga.com> In-Reply-To: <199701031713.SAA00545@yedi.iaf.nl> from "Wilko Bulte" at Jan 3, 97 06:13:40 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> Exabytes (the older ones) in general are 'sensitive' beasts. Cleaning etc > is critical. First thing I ran after hooking it up was a brand new Sony cleaning tape. > And you have to pick a 'good' firmware version from > www.exabyte.com. The one I use was recommended to me by an Exabyte > employee. And works OK for me. Hmmmm... I haven't tried to play with the firmware rev. > > because nothing else really likes them either), and I sorts get OK results > > with tar -b 32 (16k blocks)... but any other block size and it barfs. > What barfs? Jan 1 17:09:58 mongrel /kernel: st0(ahb0:6:0): MEDIUM ERROR info:17 Jan 1 17:11:29 mongrel /kernel: st0(ahb0:6:0): MEDIUM ERROR info:18 Jan 2 20:52:54 mongrel /kernel: st0(ahb0:6:0): ILLEGAL REQUEST Jan 3 06:14:38 mongrel /kernel: st0(ahb0:6:0): ILLEGAL REQUEST Jan 3 06:42:26 mongrel /kernel: st0(ahb0:6:0): MEDIUM ERROR Jan 3 06:42:26 mongrel /kernel: st0(ahb0:6:0): MEDIUM ERROR info:8 Jan 3 06:42:51 mongrel /kernel: st0(ahb0:6:0): MEDIUM ERROR info:8 Jan 3 06:44:25 mongrel /kernel: st0(ahb0:6:0): MEDIUM ERROR Jan 3 06:44:26 mongrel /kernel: st0(ahb0:6:0): MEDIUM ERROR info:10 Jan 3 06:44:43 mongrel /kernel: st0(ahb0:6:0): MEDIUM ERROR info:10 Jan 3 08:26:39 mongrel /kernel: st0(ahb0:6:0): ILLEGAL REQUEST All those MEDIUM ERRRORS are on tars and dumps at other than 16k blocks. A tar on the same tape after rewinding had no problem. Sometimes I had to "mt -f /dev/nrst0 rewind" a couple of times, but it didn't log any messages, just displayed "IO Error". > The native block size is 1kB but that's the only thing to keep in mind. > And the 32kB physio() limitation of course. BTW I normally use 'dump', > not 'tar'. But 'tar' does work also. Dump, no matter what block size I select, "sees" an end of tape immediately on doing the first write. At least it hits the tape, writes nothing, and asks for the next volume. We have another 8200 here at work that has been giving similar errors under FreeBSD on a 2940.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199701031852.MAA13083>