Author Topic: DMESG messages Question  (Read 2298 times)

Offline Was_Just19

  • Hero Member
  • *****
  • Posts: 6852
  • MLU
DMESG messages Question
« on: May 01, 2011, 11:00:34 AM »

Could I please ask for somments on the last few messages from dmesg after boot up?

I have no idea if any or all of those messages have any significance, particularly the reference to barriers,  so your help would be appreciated in decyphering them.

Thanks  ;)

Code: [Select]
eth0: no IPv6 routers present
EXT3-fs: barriers not enabled
kjournald starting.  Commit interval 5 seconds
EXT3-fs (sdb7): using internal journal
EXT3-fs (sdb7): mounted filesystem with ordered data mode
EXT3-fs: barriers not enabled
kjournald starting.  Commit interval 5 seconds
EXT3-fs (sdb5): using internal journal
EXT3-fs (sdb5): mounted filesystem with ordered data mode
EXT3-fs: barriers not enabled
kjournald starting.  Commit interval 5 seconds
EXT3-fs (sdb6): using internal journal
EXT3-fs (sdb6): mounted filesystem with ordered data mode
EXT3-fs: barriers not enabled
kjournald starting.  Commit interval 5 seconds
EXT3-fs (sda9): using internal journal
EXT3-fs (sda9): mounted filesystem with ordered data mode
EXT4-fs (sda8): mounted filesystem with ordered data mode. Opts: (null)
7:3:1: cannot get freq at ep 0x84
7:3:1: cannot get freq at ep 0x84
7:3:1: cannot get freq at ep 0x84

Offline AS

  • Hero Member
  • *****
  • Posts: 4111
  • Have a nice ... night!
Re: DMESG messages Question
« Reply #1 on: May 01, 2011, 11:28:14 AM »
So, the question mostly is what are barriers ?

Speaking of journaled filesystems, if you really want to grant the integrity and consistency of the data, the kernel must be able to perform
some write operation before some other write operation. The problem that arise is that practically every hard disk now days sports some
form of write cache that could interfere with the sequence of writes. To avoid this problem some operation has been introduced (barriers)
meaning that all write operation before one specific point should occurs before the write operations after the barrier.

But .. not all hard disks look like to support the "barriers", hence the kernel message.

"Ordered data mode": continuing to speak of journaled filesystem, journaling all write operation may result in performance problems,
mostly because each write operation will result in a double write, once on journal, and once on a definitive location.

An alternate method of journaling is the so called "ordered data mode", this method really perform journalling only related to metadata,
and not the on the data itself, resulting in a good consistency but not as strong as a fully journaled filesystem, a tradeoff between
consistency and performance.

Commit interval is the maximum time the kernel will wait before to flush the cache to the hard disk.
(laptop-mode often increase this delay, resulting in power saving at price of probability to lost more data).

AS

Offline Was_Just19

  • Hero Member
  • *****
  • Posts: 6852
  • MLU
Re: DMESG messages Question
« Reply #2 on: May 01, 2011, 11:35:53 AM »
Thank you AS for the great explanation  ;)

Those messages are relatively new, so I guess are as a result of installing a new kernel.


regards

Offline AS

  • Hero Member
  • *****
  • Posts: 4111
  • Have a nice ... night!
Re: DMESG messages Question
« Reply #3 on: May 01, 2011, 12:56:16 PM »
some additional info about barriers, journaling mode, and commit delay can be found on man page mount, in the section related to ext3 options, (i.e. barriers are disabled by default for ext3 ... this was not so some time ago... or my mem is going to fail  :D  ).

AS