Re: Block device throttling [Re: Distributed storage.]

On Mon, Aug 13, 2007 at 05:18:14AM -0700, Daniel Phillips (phillips@phunq.net) wrote:

quoted

If limit is for 
1gb of pending block io, and system has for example 2gbs of ram (or
any other resonable parameters), then there is no way we can deadlock
in allocation, since it will not force page reclaim mechanism.

The problem is that sk_alloc (called from our block driver via 
socket->write) would recurse into shrink_pages, which recursively 
submits IO to our block driver and blocks on the throttle.  Subtle 
indeed, and yet another demonstration of why vm recursion is a Bad 
Thing.

I will find a traceback for you tomorrow, which makes this deadlock much 
clearer.

I see how it can happen, but device throttling is a solution we are
trying to complete, which main aim _is_ to remove this problem.

Lower per-device limit, so that the rest of the RAM allowed to
allocate all needed data structures in the network path.
Above example just has 1gb of ram, which should be enough for skbs, if
it is not, decrease limit to 500 mb and so on, until weighted load of
the system allows to always have a forward progress.

-- 
	Evgeniy Polyakov