Return to BSD News archive
Newsgroups: comp.unix.bsd Path: sserve!manuel!munnari.oz.au!uunet!cs.utexas.edu!torn!watserv2.uwaterloo.ca!watserv1!mks.com!fredw From: fredw@mks.com (Fred Walter) Subject: Re: stock 0.1 swap not enough. Message-ID: <1992Aug9.175542.16722@mks.com> Keywords: swap Organization: Mortice Kern Systems Inc., Waterloo, Ontario, CANADA References: <BsFMoo.EvF@obiwan.uucp> <1992Aug4.162010.16364@mks.com> <BsIGDv.yo@chinet.chi.il.us> Date: Sun, 9 Aug 1992 17:55:42 GMT Lines: 49 randy@chinet.chi.il.us (Randy Suess) writes: >fredw@mks.com (Fred Walter) writes: >>No, you'll have to re-install. What I plan on doing (once my system is >>more stable) is backing up my system with cpio, then using the Fixit disk > > I would suggest using tar, as cpio seem quite flakey in reading > the tape back in. Also, make sure you use wt0, not rwt0. Yes, I noticed cpio is flakey reading, but only if the tape was created with : cpio -oc > /dev/rwt0 or cpio -oc > /dev/rwt0 But if I create the tape with : cpio -oc | dd of=/dev/rwt0 bs=512k then I can read the contents back. But I always make two copies (on two different tapes) and I do a 'cpio -it' on them to make sure those tapes actually were written. Hmmm. Does the above (requirement of using 'dd') constitute a bug ? Unfortunately the above only seems to work with the 60MEG or less tapes. My attempts at backing up to 150MEG tapes seem to crap out after about 40-45 meg. (And, yes, I am using my 150 meg tape drive). One problem with using tar is that it does *not* save the /dev/* whereas cpio does. >>(and an editted /etc/disktab on the Fixit disk that has a correct swap size >>entry) to re-disklabel my primary hard disk, re-format my partitions and then >>restore from tape. > Also, disklabel -e is a neat way to modify the disk. No > need to edit /etc/disktab. I've tried using disklabel -e, and whenever I rebooted the old contents came back. I even tried using that option (-r ?) that 'makes' it modify the disk directly... and it didn't. BTW, I've seen some people mention that the swap partition must start on a cylinder boundary. Why ? fred -- Disclaimer: everything I write is my *personal* opinion and does not represent or reflect the opinion of the company which employs me.