Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Mon, Jan 12, 2009 at 04:30:03PM +0100, Stefan Seifert wrote:
> On Monday 12 January 2009 16:21:31 Lars Ellenberg wrote:
> > On Mon, Jan 12, 2009 at 03:58:17PM +0100, Lars Täuber wrote:
> > > syncer {
> > > rate 1150M;
> > > }
> >
> > your sync-rate is set _much_ too high. unit is _bytes_ not bits, we are
> > storage guys. and it is meant to be a _throttle_. increasing it way
> > beyond physical limits does hurt, not help.
>
> Well, he said that the servers are connected via 10GigE, so the sync rate
> should be within physical limits, even though it's much above the practical
> limit.
his disks cannot cope, he gave us his hdparm -t ...
> But I'm curious: why does it hurt? A naive view would be that it won't help,
> but at least not hurt, since it would just not throttle anything. Or do you
> mean, that it will hurt normal operations which is quite understandable?
because the way it is implemented.
sort of like this:
N = sync_rate / 32 kB * 0.1 second.
that much requests (32kB each) will be requested in one batch. (ok, tried...)
so if you request much more than it can handle,
it may stumble over its own feet.
it is really meant only as a throttle.
if you don't want it to throttle,
set it _slightly_ above the physical limit.
if you set it too large, it hurts.
how much it hurts depends largely on the io backend,
it may be quite noticeable, or only measurable.
but it hurts.
--
: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com
DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
__
please don't Cc me, but send to list -- I'm subscribed