home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.admin
- Path: sparky!uunet!srvr1.engin.umich.edu!uvaarpa!murdoch!holmes.acc.Virginia.EDU!op
- From: op@holmes.acc.Virginia.EDU (Olaf Pors)
- Subject: Re: Backups and filesystem activity (was: Re: Dumping a Gigabyte)
- Message-ID: <1993Jan22.145856.27889@murdoch.acc.Virginia.EDU>
- Keywords: rc.single, automatic dumps, filesystem consistency
- Sender: usenet@murdoch.acc.Virginia.EDU
- Organization: University of Virginia
- References: <1j55eiINNk3d@BSDI.COM> <1993Jan19.135733.838@walter.cray.com> <1993Jan20.191211.22253@murdoch.acc.Virginia.EDU>
- Date: Fri, 22 Jan 1993 14:58:56 GMT
- Lines: 31
-
- >How reliable (or unreliable) are the backups made from active
- >file-systems?
-
- We back up about 90 machines daily using backups by
- inode, and have been doing daily backups by inode for
- several years. No effort is made to idle filesystems.
- We've found that occasionally backbyinode issues a
- tremendous number of "bread" errors when a filesystem is
- too busy. In these cases, either backbyinode aborts, or
- completes normally. If it aborts, we repeat the backup.
- It may take several tries to get a clean backup.
-
- We've also done a lot of restores, and have never
- to my knowledge encountered an unusable backup*.
-
- I have no similar information about "smit" backups.
-
- This is an answer based on experience, but I'd like to
- hear from anyone who has a definite answer based on knowledge
- of the backbyinode source.
-
- Olaf Pors
- Academic Computing
- University of Virginia
- op@Virginia.EDU
-
- -----------------------------------------
- * Actually, I did once upon a time find the text of
- an X11 manpage is the middle of a Sun backup. This
- was obviously not written by /etc/dump, and remains
- a mystery to this day.
-