On Tuesday, October 31, 2000 14:54:50 +0000 "Stephen C. Tweedie"
<[EMAIL PROTECTED]> wrote:

> Hi,
> 
> On Mon, Oct 30, 2000 at 09:34:27AM -0500, Chris Mason wrote:
> 
>> Yes, reiserfs could do it the same way, if we start putting a transaction
>> handle pointer in the task struct.
>> 
>> But, I thought you required some way of knowing when the transaction was
>> broken up during an operation?  You don't really need it for
>> truncate/unlink, since those are supposed to be finished, even after a
>> crash in the middle.  file writes are different though, since you could
>> end up with a write partially done that isn't included in your log.
> 
> "partially done" means the same as "included in the log" as far as
> disk updates are concerned, but yes, for writes you can't necessarily
> make the same guarantees.  In ext3's case, that implies that writes
> larger than a certain size are just not guaranteed to be atomic: they
> may be broken into smaller chunks arbitrarily, but each such chunk
> will itself be atomic and will result in a consistent filesystem.
> 

Yes, I was less than clear ;-)  I meant that a crash could result in a
fully consistent filesystem with file writes that are not reflected in the
intermezzo log.  This is not a huge issue, but it gets more interesting
when you turn off data logging (and the ext3 code that flushes new blocks
before transaction commit).  One client has a bigger file with bogus data,
and another has a smaller file with good data, and intermezzo has no log
entries to help figure out which one is right.

It seems like intermezzo could log at the start and end of each write, or
rely on an atomic write flag provided by the filesystem.  Depending on the
total size of the write, there should be cases where each one is faster
than the other.

-chris


-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [EMAIL PROTECTED]

Reply via email to