really weird filesystem problem

John T Kohl jtkohl at MIT.EDU
Thu May 2 00:53:44 AEST 1991


In article <1005 at aerodec.anu.edu.au> tridge at aerodec.anu.edu.au (Andrew Tridgell) writes:

> I have recently struck a really weird file system problem under Ultrix
> 4.0 on a DS3100. The root of the problem was a "CANNOT READ BLK :
> 567664" reported from fsck. 

>         :pg#567666:bg#8192:fg#1024:\

> Basically my questions to the net are :

> 	- what caused the problem in the first case

The symptoms you describe exactly match my experience from about 5 years
ago with a 4.2BSD filesystem.

The 4.xBSD fsck wants to read its disk blocks in groups of 4.  The
partition size you have (by default, unfortunately) is not a multiple of
four.  When reading the last two blocks, fsck gets a short read and gets
confused.

Normally this isn't a problem, EXCEPT when there's something (like a
directory block) allocated in the leftover sectors at the end of the
disk.  You probably have a fairly full partition, which ensures those
blocks get used.

> 	- how can dump/restore or tar transfer a bad block between drives?

This also explains why the problem apparently "followed" the data to the
other disks, and why it went away when you changed the partition size to
an integral multiple of 4.

> BUT (there has to be a but)
> I now have an inconsistancy between the partition table and the file
> system. The partition table thinks there is 567666 sectors, the file
> system thinks there is 567660. Could this cause a problem in the future?

It should not cause problems unless you do another newfs(8) on that
filesystem without adjusting the partition size.  You may want to use
chpt(8) to adjust the system's idea of the filesystem size to avoid this
in the future.

> 	- have I now got a time bomb waiting to go off?

Not as far as I can determine from your descriptions.
--
John Kohl <jtkohl at MIT.EDU>
Digital Equipment Corporation/Project Athena
(The above opinions are MINE.  Don't put my words in somebody else's mouth!)
___This signature printed on recycled bits___ [not original; heard 2nd hand]



More information about the Comp.unix.ultrix mailing list