->> > I'm searching for bdflush maintainer: I've created new version of
->> > bdflush (1.6.2) based on bdflush-1.5, which adds sleep
->> > support. (Bdflush will decide to spin disks down after some period of
->> > inactivity. Very usefull on notebooks.) I would like to propagate this
->> > into official tree...
->>
->> I made some updates to bdflush a while ago and sent them off to
->> belong in bdflush -- use "hdparm -S" instead.
->Well. it does belong in bdflush. For time disks are spinned down I do
->no updates, so you do not spin disks back up unless you read from
->non-cached place on disks. I also do full sync before spinning disks
->down etc. It is very convient: you can actually work on machine with
->disks spinned down. [I developed software in such state.]
Another possible use. I have an machine which doubles as firewall.
It is on an ups, but since ups can last only few hours, it would be
desirable that after ups communiate power-down, I could take
some measures to save the power.
Since this machine is just cpu/mb/ram/hdd/2*nic, I belive I could
save quite a bit of power by stopping hdd and throttling cpu
as soon as some upsd (ups daemon) signal power fail.
Still still not exactly sure how to approach this problem. Quite
bulk of the stuff belongd to upsd which would listen to
ups, turn on power-save mode of cpu (it is cyrix, so it probably
would be suspend-on-halt).
But how one should approach hdd problem? The thing is that when power
fails, the machine should basically switch info fire-wall only mode and
refuse any nfs servies. (that is any rquest to hdd, not just nfs). So, as
I see, the question is how to I tell kernel that it should *gracefully*
try that rquest later and do not bang alarm that it can't access hdd (I
don't want it spin up until power is back).
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu