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

delayed write tuning (Re: Best file system)


>>Delayed writes in XFS makes all of the above useless.  It simply doesn't
>>send the data to the disk, and that makes the potential data loss window a
>>order of magnitude (or more) bigger than the time it takes to flush data to
>>the disk with such a setup.   Although it is very possible that one can tell
>>XFS to shorten or close the delayed-write window (I didn't check).
> /proc/sys/vm.

ok, I read /usr/src/linux/Documentation and found some tuning parameters
(my comments, as I understood the docs):

/proc/sys/vm/dirty_writeback_centisecs (500)
Wakeup interval of pdflush daemon, which writes buffers to disk

/proc/sys/vm/dirty_expire_centisecs (3000)
Age threshold, after which dirty buffers get written to disk

/proc/sys/fs/xfs/xfssyncd_centisecs (3000)
xfssyncd metadata write interval

/proc/sys/fs/xfs/xfsbufd_centicecs (100)
Wakeup interval for xfsbufd metadata buffer daemon (uh, another metadata

/proc/sys/fs/xfs/age_buffer_centisecs (1500)
Max age for dirty metadata buffers

But how does this all fit together? Or, to ask the one simple question:
Whats the maximum age of new/changed data which gets lost on an xfs in
case of power failure?


CHECON   EDV-Consulting und Redaktion
         Claus Herwig * Barer Straße 70 * 80799 München
         +49 89 27826981 * Fax 27826982 * c.herwig@checon.de

Reply to: