Re: JFS in Linux [Was: Reiser4 patches.]
On Thu, Apr 03, 2008 at 08:27:48AM +0100, Steve Dobson wrote:
> On Thu, 2008-04-03 at 09:44 +1100, Dean Hamstead wrote:
> > From what i have read JFS may not be in the kernel much longer. Dont
> > know, worth someone filling me in.
>
> Now you have me worried. Were did you read this? I was thinking moving
> some of my partitions to JFS, but if it is going to be removed from the
> kernel...
I used to run JFS after Reiserfs messed my system royally. I then had
some non-specific things (e.g. where'd that file go, I know it was here
somewhere...) without errors. After asking on this list, I received an
email from the jfsutils debian package maintainer. I'll paraphrase here
since the true version will be in the archives. He sent me a copy of an
email he had received from the IBM person responsible for JFS in Linux
who said that the team was reduced from a few (don't know how many)
full-time people keeping JFS up-to-date and fixing bugs as they were
found, to just him on a half-time basis. He said that he would not run
JFS in either an existing production environment or for a new project.
Perhaps this message has percolated through to Linus. I don't know.
That thread then went into a discussion on the relative merits of
JFS/XFS/ReiserFS/ext2/ext3. The upshot was that unless you have a UPS,
don't use XFS since it handles unclean unmounts (e.g. powerfailures)
more ungracefully than the others. It was noted, however, that XFS is
great for the throughput demands of multimedia, while JFS was designed
for rapid rebooting of network servers.
Of course, over on OpenBSD they've said that the problem is inherent in
the Linux kernel and that BSD's ffs and ffs2, while needing a
comparatively long fsck time after unclean shutdown, will not lose files
the same way.
Doug.
Reply to: