Alan Cox wrote:
> No data should get more than about 30 seconds behind by default. With the
> -o sync option on ext2fs mounts it should be virtually instantaneous.
-o sync is inappropriate for slow media as it ends up not coalescing
writes, doing multiple writes to the same block, and excessively
serialising reads and writes.
What you want is to run as fast as usual, with all the nice elevator and
write back rules, but with writes to the disk sustained until there
aren't any further dirty blocks.
This strikes me as perfectly reasonable behaviour for all removable
media and I can see no reason why anyone argues against it. It's not
like there's a performance cost to simply writing your dirty data when
the medium's request queue empties.
> > Are you saying that it is not feasible to implement such a system,
> > or/and it is stupid and bad to implement such a system?
>
> We have it I think. That you saw a case it didnt work might indicate bugs
> elsewhere.
Probably he removed the drive <30 seconds after the light went off.
-- Jamie
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
