Re: Sluggish system while copying large files.

On Mon, Oct 23, 2006 at 11:40:05AM +0200, Brice Figureau wrote:

I'll try the data=writeback mount option. Is it possible to remount the
partition and change the journal option on a running system?

Not with ext3, sorry.


Since mysql is probably triggering tons of fsyncs or O_SYNC writes,
you may want to increase the size of the ext3 log.

Mysql is using O_DIRECT for its datafile. I don't know how it relates to
the sync things, but I guess that to be truly ACID, it has to fsync the
files on each transactions.
How can I increase the ext3 log ?
Any idea of the size I should use (and what is the default) ?

Even when using O_DIRECT, if the file is extended an FS commit is
triggered to record the extension.

If mysql is constantly appending to the files holding your tables, the
synchronous writes are more expensive and log intensive. Check your
setup to see if you can manually extend any of those files to avoid
constantly growing table files.

The use of a battery-backed RAID cache should mitigate the sync writes,
and since our mysql load is quite low, the machine shouldn't definitely
freeze for seconds while copying files.

Ok, I would start with only doing data=writeback. That is probably a
big part of the problem.

-chris

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Relevant Pages

  • [PATCH 2.6] ext3: fix documentation of online resizing
    ... Undocument the non-working resize= mount option in ext3, ... references to the ext2resize package instead, ... only proper way of doing online resizing of ext3 filesystems. ...
    (Linux-Kernel)
  • [PATCH RFC] Ext3 data=guarded
    ... Here is a respin of the ext3 data=guarded code. ... This adds a new mount option ... the on disk i_size after all of the data blocks are on disk. ... More details are in the patch emails. ...
    (Linux-Kernel)
  • Re: [Ext2-devel] [RFC 0/13] extents and 48bit ext3
    ... LT> major surgery on ext3. ... It's the main filesystem for a lot of users, ... mount option. ...
    (Linux-Kernel)
  • Re: [Ext2-devel] [RFC 0/13] extents and 48bit ext3
    ... LT> major surgery on ext3. ... If it will remain a mount option, if it is never made the default, then only 1% of users will ever use the feature. ... And we shouldn't merge a 1% use feature into the _main_ filesystem for Linux. ...
    (Linux-Kernel)