Re: patch: reiserfs for 2.3.49

From: Jamie Lokier (
Date: Sat Mar 18 2000 - 07:42:40 EST

I'll add

      * Make sure everything reaches the disk

I remember turning off `update' because after the equivalent was built
in to the kernel, but it was removed at the next kernel version. (It
got put back soon after).

One day, at the end of the day, the system crashed for some reason.
Unfortunately, much of the data I'd written during the _entire_ day
hadn't reached the disk because nothing was flushing it to disk.

-- Jamie

Riley Williams wrote:
> Allied to that...
> * Fault detection
> If a fault occurs on a filesystem that has the ability to
> damage a file we are using, we want to know about it as
> soon as possible. Specifically, we want to be able to
> write programs that ALWAYS read the file directly off the
> hard drive, ignoring any cache.
> I still remember one problem we had on a (thankfully non-
> Linux) system where the fact that a file descriptor had
> been leaked that pointed to the file being open for read
> meant that even a program designed to look for errors in
> that file was unable to detect any as the file was just
> read from cache each time, and NOT from the hard drive.
> As a result, the first time we became aware of any damage
> to the file was MONTHS after the event when the machine
> was taken down for an upgrade - when it came back on, we
> had to spend nearly a week just sorting out the problems
> with the early records in that file (which had last been
> WRITTEN to at least four months previously) despite the
> fact that we had been running a program DAILY to READ the
> entire file to date and verify that it was complete.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

This archive was generated by hypermail 2b29 : Thu Mar 23 2000 - 21:00:24 EST