Subject: Re: misc/1919: merging functionality of netstart, etc, into rc
To: None <netbsd-bugs@NetBSD.ORG>
From: Simon Burge <>
List: netbsd-bugs
Date: 01/11/1996 13:35:07
On Jan 10,  1:15pm, Jason Thorpe wrote:

>  > - "fsck -p" becomes "fsck -f -p".  And it stays that way at least until
>  >   clean bits really work, and probably until check-periodically-anyway
>  >   clean bits go in.
> For the record, I think that the "clean bit" actually does work now .. 
> it's even enabled in the Alpha sources again.  This was fixed sometime 
> before 1.1.  Re. "check anyway", yah, that should happen.  Maybe I'll 
> look at it sometime "soon", but others are welcome to try their hand at 
> it, too :-)
> Would folks rather see a "mount count" or "time since" check?

Why not do what Ultrix does - both?  From the fsck(8) man page:

     The -p option should be used to check file systems.  The generic file
     system interface, gfs, causes fsck to realize when a file system is
     unmounted cleanly and thus prevents fsck from doing the check.  File
     systems are unmounted cleanly only when an error-free shutdown has been
     performed or the file system was unmounted.  However, a timeout factor
     is used by fsck to determine if fsck should be run regardless of the
     value of the clean byte.  The timeout factor is initially set to 20 and
     is decremented when any one of three events occur:

          -  A file system is mounted,

          -  10,000 updates have occurred

          -  A file system was updated and fsck occurred more than 60 days

     When the timeout factor reaches 0, fsck will automatically check it.
     This factor can be changed with tunefs.  If the -P option is used, the
     parallel consistency checks are performed like the -p option regardless
     of how the file system was unmounted.