Path: utzoo!attcan!uunet!apctrc!zjat02 From: zjat02@apctrc.UUCP (Jon A. Tankersley) Newsgroups: comp.unix.questions Subject: munged root fs on SunOS 3.X (was Re: umount, and SunOS 3.X) Message-ID: <533@cra2.apctrc.UUCP> Date: 12 Sep 88 18:51:37 GMT References: <27262@neabbs.UUCP> <520@cra2.apctrc.UUCP> <13448@mimsy.UUCP> Reply-To: zjat02@cra2.UUCP (Jon A. Tankersley) Organization: Amoco Production Co, Tulsa Research Center Lines: 38 In article <13448@mimsy.UUCP> chris@mimsy.UUCP (Chris Torek) writes: >In article <520@cra2.apctrc.UUCP> zjat02@apctrc.UUCP (Jon A. Tankersley) >writes: >>rc.boot trashes /etc/mtab before running /etc/umount -at nfs. This leaves >>potential garbage on remote systems /etc/rmtab (remote mtab). > >It does more than just that. rc.boot is always run, even on a single- >user boot. If your root file system gets corrupted---and as most of us >know, anything that *can* go wrong *will* go wrong (someone at Sun >seems to have forgotten)---the code that clobbers /etc/mtab will cause >a continuous series of crashes and auto-reboots. The only way to >recover is to boot from another machine, or from the distribution >tapes. But if this is the first machine on your network, or the only >one with source . . . ? (Praise to Aule for mini-roots.) I've been able to fix this problem by -not- rebooting single user. It seems to be able to fix most of the problems in the root and reboot. I then halt that reboot and boot single user. A pain. But it works. -tank- You Quessed It -- #include /* nobody knows the trouble I .... */