Corroupted CVS File (fwd)

Cy Schubert Cy.Schubert at komquats.com
Thu Jul 3 10:22:38 PDT 2003


The fsck discovered an unexpectd softupdate inconsistency,
listing the file in question.
Case closed.


Cheers,
--
Cy Schubert <Cy.Schubert at komquats.com>        http://www.komquats.com/
BC Government                     .                       FreeBSD UNIX
Cy.Schubert at osg.gov.bc.ca         .                     cy at FreeBSD.org
http://www.gov.bc.ca/             .            http://www.FreeBSD.org/


------- Forwarded Message

Date:    Thu, 03 Jul 2003 09:22:25 -0700
From:    Cy Schubert <Cy.Schubert at komquats.com>
To:      John Polstra <jdp at polstra.com>
Subject: Re: Corroupted CVS File (fwd) 

In message <200307031611.h63GBLRf009280 at strings.polstra.com>, John 
Polstra writes:
> In article <200307031602.h63G20HN071272 at cwsys.cwsent.com>,
> Cy Schubert  <Cy.Schubert at komquats.com> wrote:
> 
> > It appears I may not be going insane after all.
> 
> I never suspected you of insanity. :-)

I know you didn't. It's just my way of putting it.  :)

> 
> > The file is received OK from cvsup7 and from cvsup10. When I check
> > out (using cvs co) from my local repo or rsync the repo to another
> > machine on my network here at home, the file is corrupted. Looks
> > like I've found some kind of issue with -CURRENT. I'm running
> > 5.1-REL.
> 
> Next time you find a corrupted RCS file, save it somewhere.

Not a problem. I can reproduce this problem at will.

>  Take a
> look at it in an editor and see if you can spot what's going on.  The
> typical kinds of corruption I've seen over the years when people have
> reported these things to me have been:
> 
>     - Single-bit errors in the RCS file.  These are always caused by
>       HW problems, and they always go away if ECC RAM is installed.

I doubt this

> 
>     - Either a region of zeroes or a portion of an unrelated file
>       splatted somewhere into the middle of the corrupted RCS file.

Possible. I'll have to look more closely

>       This kind of corruption is always page-aligned or filesystem
>       block aligned.  It is probably caused by kernel software bugs.
>       I used to see a lot of these back around FreeBSD-3.x, but I
>       haven't heard of any for a long time now.

Interestingly, my repo exists on my /opt2 filesystem and /opt2 appears 
somewhere in the middle of the file. It's the only intelligible string 
in the file. Could I have some kind of single bit error (h/w) or 
storage overlay (h/w or s/w)? I guess I'm stating the obvious. I 
suspect what I'm seeing is either random data on disk because when the 
inode is rewritten to update the access time it's corrupted.... 
hmmm.... could I have some filesystem corruption?? I think I'll shut 
down and force an fsck.

> 
> John
> -- 
>   John Polstra
>   John D. Polstra & Co., Inc.                        Seattle, Washington USA
>   "Two buttocks cannot avoid friction."                     -- Malawi saying



Cheers,
- --
Cy Schubert <Cy.Schubert at komquats.com>        http://www.komquats.com/
BC Government                     .                       FreeBSD UNIX
Cy.Schubert at osg.gov.bc.ca         .                     cy at FreeBSD.org
http://www.gov.bc.ca/             .            http://www.FreeBSD.org/




------- End of Forwarded Message



More information about the freebsd-current mailing list