Announcement

Collapse
No announcement yet.

EXT4 Data Corruption Bug Hits Stable Linux Kernels

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #51
    Originally posted by tytso View Post
    Please send a detailed report to the linux-ext4 mailing list. Include the kernel version you are running, the kernel logs (from dmesg) that were printed or which you found in the logs, the output from e2fsck (which are hopefully saved in /var/log/fsck), and whether you can reliability reproduce it or how frequently it happens.

    There are many potential causes of file system corruption, and sometimes they can have very similar symptoms. This is why I hate try to debug such problems on web forums or on Ubuntu's Launchpad. Someone sees a relatively vague description of the problem on the web forum, immediately jump to the conclusion that it must be the same thing that they saw, and pile on the web thread. Always send us detailed information; if it's redundant with what someone else has sent, that's actually *good*. That way we can confirm a common pattern. But please don't assume a pattern where one doesn't exist; humans tend to do this even when it's not justified.
    i did it twice with the same outcome. First time i guessed it's hardware problem, so i replaced sata cables and went with some memory and hdd tests, which went fine. Afterwards re-installed system with same amount of reboots, and got file system corruption again. Finally excluded kernel* from updates, keeping 3.3.4-5.fc17.x86_64, with no problems since. I'll do everything again to get some logs for you.

    Comment


    • #52
      thanks ryao, I've actually been using the latest hardened gentoo kernel in a laptop that I use for political activism. Only good things to say about it.


      I'm fairly new to linux but all the articles that I've red stated clearly that ZFS was the best thing to come out of oracle... light years ahead of the major file systems and anyone could see it was the superior FS, even here I saw multiple posts stating the same.


      I know that mass migration to ZFS won't happen but how far can you take this EXT thing? EXT7?

      instead of pricking a dying horse (with the consequences exposed here) I would request all distros to give you ZFS as an option

      Comment


      • #53
        @Pallidus

        I recall there are some issues with licensing and patents, and that kind of stuff ..
        Here's some related info.
        I haven't used ZFS myself though, ext4 seems enough for me. I assume people are talking about the original ZFS and not any "ports" under development?

        Comment


        • #54
          Originally posted by frantaylor View Post
          Apparently Linus Torvalds is actually just a bot that collects [... the creaking continues ...].
          Apparently frantaylor is just this whiny little child who spams around the internet, running his mouth off without putting an inch of thought into anything because he's unable to do so with the smelly mush that was planted into his head. This detection of sorts is something called "beyond of reach" to him as his gloomy eyes are unable to detect the well-known fact that Greg, not Linus, G-R-E-G is the one who sings-off on and maintains the stable branch. It is somehow saddening, this patheticism in just a one single person. One can only hope this disease will not spread and no female will comply and decide to reproduce with him. It will be the end of our species as we know it.

          Look, and feast your feeble brains with something you will never understand anyw... actually forget about everything I ss... no wait you wouldn't understand any of that either! So (from the changelog of 3.6.2 - those are called numbers.. can you count all the way up to 6?):

          jbd2: don't write superblock when if its empty commit eeecef0af5ea4efd763c9554cf2bd80fc4a0efd3 upstream.
          [...]
          Signed-off-by: Eric Sandeen <[email protected]>
          Signed-off-by: "Theodore Ts'o" <[email protected]>
          Signed-off-by: Greg-not-Linus Kroah-not-Torvalds-Hartman <[email protected]>
          So, as Paradox Uncreated once so elegantly put it:
          Originally posted by Paradox Uncreated View Post
          Fuck you in the biggest way.

          Comment


          • #55
            Originally posted by Rigaldo View Post
            @Pallidus

            I recall there are some issues with licensing and patents, and that kind of stuff ..
            Here's some related info.
            I haven't used ZFS myself though, ext4 seems enough for me. I assume people are talking about the original ZFS and not any "ports" under development?
            I did a write up for the Sabayon community about this. in short, the stuff at Wikipedia is inaccurate, I need to find time to correct wiki pages and using ZFS in the form of kernel modules is fine. You can read my post on the Sabayon forums for more details.

            Comment


            • #56
              ZFS

              I have been using ZFS for a couple of months on boxes that present iSCSI storage off to VMWare hosts (set up a huge RAIDZ2 array, then make 2TB zvols in it, share those as iSCSI LUNs...", and tried it out specifically to get away from using hardware RAID cards that take forever and a week to rebuild a 10TB array. It seems to pretty much "just work", and I was able to train one of our mid-level tech minions how to do the setup and config. That said, I don't really feel any burning need for it on smaller boxes or desktops.

              Comment


              • #57
                Originally posted by ryao View Post
                I did a write up for the Sabayon community about this. in short, the stuff at Wikipedia is inaccurate, I need to find time to correct wiki pages and using ZFS in the form of kernel modules is fine. You can read my post on the Sabayon forums for more details.

                https://forum.sabayon.org/viewtopic....it=ZFS#p154165
                Cool, thanks for the heads up.

                Comment


                • #58
                  Originally posted by ArchLinux View Post
                  well-known fact that Greg, not Linus, G-R-E-G is the one who sings-off on and maintains the stable branch.
                  It was a very,very special and unreasonable setup which triggerd the data loss for just two people.

                  Comment


                  • #59
                    Who is responsible for blowing this out of proportions?

                    Note:  This blog post outlines upcoming changes to Google Currents for Workspace users. For information on the previous deprecation of Googl...



                    so default mount options were safe

                    Comment


                    • #60
                      Originally posted by ulenrich View Post
                      It was a very,very special and unreasonable setup which triggerd the data loss for just two people.
                      I resemble that remark. It wasn't an 'unreasonable' configuration per se, since the documentation doesn't actually mention that journal_async_commit is experimental. Nowadays it mentions that it enables journal_checksum, but it doesn't mention that that option is experimental let alone dangerous, and years back when I set up the /etc/fstab on the failing machine journal_checksum had never caused anyone problems in practice (this changed a couple of years later). And it does speed things up a good bit on metadata-heavy workloads, which I cared about.

                      And then I used that configuration for three and a half years with not a single tiny problem. Sorry, I don't generally go back to three-and-a-half year old working system configurations and wonder if perhaps they are too dangerous and should be changed! -- not until they actually cause problems, that is.

                      Now if you use nobarrier without battery-backed hardware RAID or something equally high-end with independent power (like a SAN), you are indeed doing something unreasonable, verging on suicidal. But battery-backed hardware RAID isn't actually that expensive anymore, and isn't that rare: it added less than ?100 to the cost of this machine when I bought it (and performance-wise, it flies!).

                      And, hey, even if it was unreasonable, part of the reason why I run with strange options is so that I can report bugs in them

                      What was probably unreasonable was splashing a conversation/debugging-in-progress all over the media. I still haven't found all the places this got published, almost always hyped-up beyond any reasonable degree, and with the exception of LWN (where I posted a 'hey, hold off upgrading for now' comment) *everyone* got it wrong. Oh well. (As far as I can tell, Phoronix was first, and was probably the source from which this fountain of inaccuracy sprang.)

                      Comment

                      Working...
                      X