Date: Mon, 26 Apr 2004 00:34:55 -0500 From: "Richard J. Valenta" <RJV@WEBLINKMO.COM> To: <freebsd-alpha@freebsd.org>, <freebsd-questions@freebsd.org> Subject: odd 4.9 scsi problem Message-ID: <5EEBE9C3C61D1142994C6B620C51E8470CC842@ip-64-32-234-4.dsl.iad.megapath.net>
next in thread | raw e-mail | index | archive | help
I have an odd scsi problem - this is a DS20 with 3 Qlogic scsi cards, 0 has only the external TZ89 DLT drive, 1 has the hard drives, 2 has the internal DAT (TLZ,T1Z10)and the cdrom. If I power down all devices completely including the server (not just reboot) and then power everything up, this is the result: worx# camcontrol devlist <DEC TZ89 (C) DEC 2150> at scbus0 target 3 lun 0 (pass0,sa0) <DEC RZ1CF-CF (C) DEC 1614> at scbus1 target 0 lun 0 (pass1,da0) <COMPAQ BB00911CA0 3B05> at scbus1 target 1 lun 0 (pass2,da1) <COMPAQ BB00911CA0 3B05> at scbus1 target 2 lun 0 (pass3,da2) <DeC TlZ10 (C) dEc 04a8> at scbus2 target 4 lun 0 (pass4,sa1) <DEC RRD47 (C) DEC 1206> at scbus2 target 5 lun 0 (pass5,cd0) But an odd thing happens to the DAT after I manually reset and rescan - it changes names, worx# camcontrol reset 2 Reset of bus 2 was successful worx# camcontrol rescan 2 Re-scan of bus 2 was successful worx# camcontrol devlist <DEC TZ89 (C) DEC 2150> at scbus0 target 3 lun 0 (pass0,sa0) <DEC RZ1CF-CF (C) DEC 1614> at scbus1 target 0 lun 0 (pass1,da0) <COMPAQ BB00911CA0 3B05> at scbus1 target 1 lun 0 (pass2,da1) <COMPAQ BB00911CA0 3B05> at scbus1 target 2 lun 0 (pass3,da2) <DEC TLZ\0210\000 \000 \010C\011 DEC 0\024a\030> at scbus2 target 4 lun 0 (sa1,pass4) <DEC RRD47 (C) DEC 1206> at scbus2 target 5 lun 0 (pass5,cd0) But this isn't the main problem - I just discovered this playing around after the main problem - when I try to use sa0 (the DLT), I often get an error like below: <snip tar info> archive/backup/images/Videos/rich.mov tar (child): /dev/sa0: Wrote only 0 of 10240 bytes tar (child): Error is not recoverable: exiting now Broken pipe worx# when I try to perform the backup again it hangs and the tail of /var/log/messages shows: Apr 25 18:07:11 worx /kernel: (sa0:isp0:0:3:0): failed to write terminating filemark(s) Apr 25 18:07:11 worx /kernel: (sa0:isp0:0:3:0): tape is now frozen- use an OFFLINE, REWIND or MTEOM command to clear this state. And when I try to do so: worx# mt -f /dev/sa0 rewind mt: /dev/sa0: Device not configured and when I reboot... worx# camcontrol devlist <DEC RZ1CF-CF (C) DEC 1614> at scbus1 target 0 lun 0 (pass0,da0) <COMPAQ BB00911CA0 3B05> at scbus1 target 1 lun 0 (pass1,da1) <COMPAQ BB00911CA0 3B05> at scbus1 target 2 lun 0 (pass2,da2) <DeC TlZ10 (C) dEc 04a8> at scbus2 target 4 lun 0 (pass3,sa0) <DEC RRD47 (C) DEC 1206> at scbus2 target 5 lun 0 (pass4,cd0) Odd change of CAPS for the DAT, and the DLT has disappeared... System info: single 500mhz alpha processor DS20, with 3 Qlogic scsi cards and a brand new kernel, which only includes Qlogic cards in its config, and all scsi device types as well. I had 5.0 on this machine and ran backups fine. Then changed to 4.9 trying to get sound to work, that was unsuccessful, and now I have this tape problem, I don't know if they are related. At times, if I simply let the machine sit, the DLT will just fall off the chain and a rescan/reset will not bring it back. At the same time the other tape gets renamed. I'd like to blame a card to the tape drive, but I've moved things around and had the same problems, and would think that if the DLT was fried it just wouldn't work - not work some of the time. Love to hear anyone's opinions or thoughts on repairing this. Richard=20
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5EEBE9C3C61D1142994C6B620C51E8470CC842>