Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!att!pacbell.com!ames!bionet!agate!ucbvax!tdisys.UUCP!mogenix
From: mogenix@tdisys.UUCP (MOGENET)
Newsgroups: comp.sys.sgi
Subject: (none)
Message-ID: <9012280901.AA20329@tdisys.uucp>
Date: 28 Dec 90 11:01:29 GMT
Sender: usenet@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 34


Help !!!!

I got another problem with Silicon <--> IBM RS/6000 :

When I am performing a "tar cv bigfile" on my IBM RS/6000 with "bigfile"
being around 40 Meg, and try a "tar xvof /dev/tapens" on my PI 4D/25,
here's what happens:

	Around 7.7 Meg of "bigfile" are extracted, then tar asks me
	to change tape. Well, as I can't change tape, I tried
	"mt rewind; tar xvof /dev/nrtapens" and when it asks me to
	change tape, I just extract it and put it back. As you guessed,
	it didn't work either. 

	I wrote then some piece of code performing a per-block reading
	of /dev/tapens, and which asks the driver status when the read(2)
	return -1 via some MTIOCGET ioctl.

	That marvelous ioctl tells me that:

		1. Streamer is a Western Digital SCSI
		2. That it is on-line
		3. That it has reached an EOD condition (not EOT)

Could someone around involved in the subtleties of SCSI Streamers tell
me if there is a solution ?

		Thanks,

				Mogenix![120q

P.S.: No need to say that I broke through by cutting "bigfile" in under-7 Meg
pieces, but let me tell you, 't ain't no fun.