Ronny Buchmann wrote:

> * Hans Reiser <[EMAIL PROTECTED]> [2001-07-31 18:44] schrieb:
> 
>> David Rees wrote:
>> 
>>> On Tue, Jul 31, 2001 at 03:06:14PM +0400, Hans Reiser wrote:
>>> 
>>>> So a RedHat user has to:
>>> 
> a redhat user with reiserfs as /
> 
>>> I wouldn't say HAS to, but...
>>> 
>>>> change the halt script to unmount reiserfs before cutting power
>>> 
>>> Definately a good idea, but I've been running for ages without it without
>>> any problems.
>> 
>> Correct me if I am wrong, but if you were to write a 1gb file, then type halt,
>> you could get a corrupted file instead of a perfect one, yes?
> 
> all processes are killed before the real halt, so this can't happen i think
> 
>>>> recompile using the latest kernel
>>> 
>>> Also definately a good idea to get all the latest bug fixes.
>>> 
>>>> avoid using a gcc not recommended by the kernel README
>>> 
>>> If you upgrade all your RedHat errata you should be fine.  I've been using
>>> that for all my reiserfs machines.
>> 
> ACK
> 
>>>> change the mount scripts to not run fsck on reiserfs.
>>> 
>>> Or `ln -s /bin/true /sbin/fsck.reiserfs`.  Otherwise on my various RedHat
>>> machines I get message that says "Warning, fsck.reiserfs not found" or
>>> similar unless you install the reiserfs-utils package in RH 7.1 where
>>> you'll definately want to do this.
>> 
> if one uses reiserfs he should install reiserfs-utils package (if one wants beta 
>stuff, he can build new beta rpm)
> fsck.reiserfs is a link to reiserfsck and it's always called with -a (or -p, not 
>100% sure what fsck actually does)
> 
>>>> Anything else?
>>> 
>>> I think that's it!  Anyone want check to see if these issues are fixed in
>>> Rawhide, and if not, open bugs in the RedHat bug db?  I can do it, I just
>>> may not have time for a little bit.
>>> 
>>> -Dave
>> 
>> If someone can do that it would be nice.  Productive.  Better than our updating
>> the faq.
> 
> so i think there are no issues, maybe the remount,ro thing on halt
> what is the right solution here (also for other fs like xfs,  nfs and so on)
> 
> currently redhat only remounts ext2 read-only
> 
> maybe i should file a bug report to see what the redhat guys think
> 
>> Hans
>> 
Killing  a process does not move data out of  kernel cache and onto disk.

Hans

Reply via email to