Quick Search | Power Search | Search Filter | Interest Finder | Browse Groups | Post Message |
Article 2 of 5 | Text Only Help |
Previous Article |
Next Article |
Current Results |
View Thread |
Post Message |
Subject: Re: Map SCSI block# to file From: ben@REX.RE.ouhsc.edu (Benjamin Z. Goldsteen) Date: 1997/08/06 Message-Id: <5sb0re$6l2@REX.RE.ouhsc.edu> Newsgroups: comp.sys.sgi.admin [More Headers] ben@REX.RE.ouhsc.edu (Benjamin Z. Goldsteen) writes: >I made the unfortunate decision not to thorougly test my new Elite 23 before >migrating files over to it (I assumed bad blocks, etc were already taken >care of). Now it is reporting many errors -- several unrecoverable. I >would like to map the SCSI block # in the SYSLOG to the file so that I may >restore just those files from backup. Sorry, I didn't mention that this is on XFS file system. Anyway, I found a solution: I took the addresses from SYSLOG and then used xfs_db (which has no man page and isn't documented anywhere): blockget -n daddr block# blockuse ncheck -i inode# I tried to automate this: echo << EOF | xfs_db ... daddr 1234 blockget blockuse EOF But xfs_db returned no output. This is rather unfortunate since I had about 40 or 50 addresses I needed to lookup. I created a file that I only had to cut and paste into a window running xfs_db, but I still would have prefered to do this from the command-line. Another frustruation is that xfs_check doesn't work as advertised. It claims to take "-b fsblock#" as an argument but it doesn't. One can just view the shell script and see that it doesn't. I have patch 1768 installed. A few more notes from this morning: I tried to use fx/execercise to map out a bad block and it wouldn't. It found all sorts of errors with this block but never marked it as bad. I had to do it manually. Not the worst thing, but I don't understand why it didn't work. Also, the messages in the SYSLOG only list the block number (which I assume is the absolute block number). The manual pages suggest that two numbers should be listed for SCSI disks: one absolute to the beginning of the disk and one relative to the partition. By the way, is this bad: dks1d4vol: Recovered Error: Defect list not found (asc=0x1c, asq=0x0), (cmd byte 2) (bit 2) ? I am going to do a full exercise/verify this weekend... -- Benjamin Z. Goldsteen
Previous | Next | Results | View Thread | Author Profile | Post Message | Post Reply | Send Email |
Copyright © 1995-97 Deja News, Inc. All rights reserved. |