rjg@sq.com (Bob Gibson) (10/08/90)
Machine: Sun 4/330 OS: SunOS 4.0.3 Disk: CDC Wren IV (94171-344) Controller: Internal SCSI The problem: We have a bad block in the /usr partition of this disk. The "repair" facility of format(8) wants to read the "current defect list" from the disk. This was a replacement disk from Sun. For some reason, it doesn't contain one. The System Administration Manual outlines the following three methods for (re)creating a defect list: (1) "load" one from a file that was previously "dump"ed to (2) "extract" one from the disk (3) enter the defects manually with "add" Since this disk never contained a defect list, one was never "dump"ed. My attempt to "extract" one from the disk resulted in the message "Extraction failed" with no explanation. This leaves me with door number (3). The "original" command reports 45 original bad blocks, but I assume I can't create my defect list soley from these since there may have been more encountered during the original format. I figure I have to use "analyze" to find them all, "add" to build the working list, "commit" to set "current" from "working", and "backup" to write "current" to disk. Is this correct? The "commit" command warns that the disk must be reformatted for the changes to take effect. The manual claims that individual portions of the disk can be formatted so presumably I can reformat only the bad spots. Is it necessary to reformat for all of them, or just the new one? I would like to avoid reformatting the entire partition if possible since restoring it will be a pain. On a related note, is it reasonably safe to do the format in single user mode, or do I have to do it across the net via MUNIX? Anyone care to share their wisdom on this? I have the bad block isolated in a file, but dump(8) won't let me dump this filesystem until I take care of this. Thanks in advance, Bob Gibson Internet: rjg@sq.com SoftQuad Inc. UUCP: ...!utzoo!sq!rjg