[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: when does journal flush with data=journal



Theodore Tso wrote:
> 
> On Fri, Feb 08, 2002 at 12:59:50PM -0800, Andrew Morton wrote:
> > The other (complimentary) option is to simply hack mke2fs so that it permits
> > larger journals.  In e2fsprogs's misc/util.c:figure_journal_size():
> 
> Note though that interactivity may suffer with larger journals; when
> it finally *does* come time to flush the journal, the system may
> become very non-responsive until the journal is actually completely
> flushed.

Yes.  Once we get into forcing checkpoints, new writers can get
*seriously* gummed up, because the filesystem is actively writing
things to the main disk (ie: very seek-intensive), which is much
slower than journal writes.

And increasing the journal size will increase the amount of data which
is checkpointed, when a checkpoint force happens.

Neil Brown found that his NFS server was getting bitten by this -
pauses of a few seconds every thirty seconds, I think.  His workaround
was to drastically shorten the kupdate interval so that kupdate
performs the nice smooth checkpointing for us.  This is discussed
in Daniel Robbins' second article, at 

	http://www-106.ibm.com/developerworks/linux/library/l-fs8/

ext3's checkpointing code is too drastic at present - too much, too
late, too infrequent.  We do need a native solution, rather than
borrowing kupdate functionality, I guess.

-





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]