On Mon, 2005-10-10 at 20:57 +0100, Peter Grandi wrote:
> I have been a fairly happy user of JFS for the past few weeks
> and having looked at it and other filesystem I still have a few
> questions:
> 
> * Is it still the case that as reported in 'man'(8) JFS does not
>   implement the 'sync' mounting option as the «sync option today
>   has effect only for ext2, ext3 and ufs»?

There might be holes, but for the most part the sync option is handled
by the generic vfs code.  The code in mm/filemap.c checks the
MS_SYNCHRONOUS flag and does the right thing.

> * What regulates the journal commit interval? With 'ext3' one
>   can explicitly regulate it at mount time with '-O commit=N',
>   is there something similar?

jfs doesn't specify its own interval.  It relies on pdflush to ensure
that data makes it to disk in a reasonable time.

> * If JFS does not have its own commit interval parameter, are
>   the journal blocks subject only to the settings in the various
>   '/proc/sys/vm/dirty_*' parameters, and if not, what is the
>   policy for syn'ing updated journal blocks?

Yes.

> * Is 'jfs_fsck' multithreaded? If not, how well or badly does it
>   scale to RAIDs with many large discs, say 10-20 250GiB ones?

No it isn't.  I'm afraid that scalability is a problem with jfs_fsck.
It's not so much dependent on the size of the volume, but the number of
inodes.  In other words, files that contain large extents of contiguous
disk blocks won't hamper jfs_fsck, but having a very large number of
files & directories can cause problems.
-- 
David Kleikamp
IBM Linux Technology Center



-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
_______________________________________________
Jfs-discussion mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jfs-discussion

Reply via email to