Hello Fengguang, On Fri 25-03-11 21:44:11, Wu Fengguang wrote: > On Wed, Mar 23, 2011 at 05:43:14AM +0800, Jan Kara wrote: > > Hello Fengguang, > > > > On Fri 18-03-11 22:30:01, Wu Fengguang wrote: > > > On Wed, Mar 09, 2011 at 06:31:10AM +0800, Jan Kara wrote: > > > > > > > > Hello, > > > > > > > > I'm posting second version of my IO-less balance_dirty_pages() patches. This > > > > is alternative approach to Fengguang's patches - much simpler I believe (only > > > > 300 lines added) - but obviously I does not provide so sophisticated control. > > > > > > Well, it may be too early to claim "simplicity" as an advantage, until > > > you achieve the following performance/feature comparability (most of > > > them are not optional ones). AFAICS this work is kind of heavy lifting > > > that will consume a lot of time and attention. You'd better find some > > > more fundamental needs before go on the reworking. > > > > > > (1) latency > > > (2) fairness > > > (3) smoothness > > > (4) scalability > > > (5) per-task IO controller > > > (6) per-cgroup IO controller (TBD) > > > (7) free combinations of per-task/per-cgroup and bandwidth/priority controllers > > > (8) think time compensation > > > (9) backed by both theory and tests > > > (10) adapt pause time up on 100+ dirtiers > > > (11) adapt pause time down on low dirty pages > > > (12) adapt to new dirty threshold/goal > > > (13) safeguard against dirty exceeding > > > (14) safeguard against device queue underflow > > I think this is a misunderstanding of my goals ;). My main goal is to > > explore, how far we can get with a relatively simple approach to IO-less > > balance_dirty_pages(). I guess what I have is better than the current > > balance_dirty_pages() but it sure does not even try to provide all the > > features you try to provide. > > OK. > > > I'm thinking about tweaking ratelimiting logic to reduce latencies in some > > tests, possibly add compensation when we waited for too long in > > balance_dirty_pages() (e.g. because of bumpy IO completion) but that's > > about it... > > > > Basically I do this so that we can compare and decide whether what my > > simple approach offers is OK or whether we want some more complex solution > > like your patches... > > Yeah, now both results are on the website. Let's see whether they are > acceptable for others. Yes. BTW, I think we'll discuss this at LSF so it would be beneficial if we both prepared a fairly short explanation of our algorithm and some summary of the measured results. I think it would be good to keep each of us below 5 minutes so that we don't bore the audience - people will ask for details where they are interested... What do you think? I'll try to run also your patches on my setup to see how they work :) V6 from your website is the latest version, isn't it? > > > > The basic idea (implemented in the third patch) is that processes throttled > > > > in balance_dirty_pages() wait for enough IO to complete. The waiting is > > > > implemented as follows: Whenever we decide to throttle a task in > > > > balance_dirty_pages(), task adds itself to a list of tasks that are throttled > > > > against that bdi and goes to sleep waiting to receive specified amount of page > > > > IO completions. Once in a while (currently HZ/10, in patch 5 the interval is > > > > autotuned based on observed IO rate), accumulated page IO completions are > > > > distributed equally among waiting tasks. > > > > > > > > This waiting scheme has been chosen so that waiting time in > > > > balance_dirty_pages() is proportional to > > > > number_waited_pages * number_of_waiters. > > > > In particular it does not depend on the total number of pages being waited for, > > > > thus providing possibly a fairer results. > > > > > > When there comes no IO completion in 1 second (normal in NFS), the > > > tasks will all get stuck. It is fixable based on your v2 code base > > > (detailed below), however will likely bring the same level of > > > complexity as the base bandwidth solution. > > I have some plans how to account for bumpy IO completion when we wait for > > a long time and then get completion of much more IO than we actually need. > > But in case where processes use all the bandwidth and the latency of the > > device is high, sure they take the penalty and have to wait for a long time > > in balance_dirty_pages(). > > No, I don't think it's good to block for long time in > balance_dirty_pages(). This seems to be our biggest branch point. I agree we should not block for several seconds under normal load but when something insane like 1000 dds is running, I don't think it's a big problem :) And actually the NFS traces you pointed to originally seem to be different problem, in fact not directly related to what balance_dirty_pages() does... And with local filesystem the results seem to be reasonable (although there are some longer sleeps in your JBOD measurements I don't understand yet). > > > > The results for different bandwidths fio load is interesting. There are 8 > > > > threads dirtying pages at 1,2,4,..,128 MB/s rate. Due to different task > > > > bdi dirty limits, what happens is that three most aggresive tasks get > > > > throttled so they end up at bandwidths 24, 26, and 30 MB/s and the lighter > > > > dirtiers run unthrottled. > > > > > > The base bandwidth based throttling can do better and provide almost > > > perfect fairness, because all tasks writing to one bdi derive their > > > own throttle bandwidth based on the same per-bdi base bandwidth. So > > > the heavier dirtiers will converge to equal dirty rate and weight. > > So what do you consider a perfect fairness in this case and are you sure > > it is desirable? I was thinking about this and I'm not sure... > > Perfect fairness could be 1, 2, 4, 8, N, N, N MB/s, where > > N = (write_bandwidth - 1 - 2 - 4 - 8) / 3. > > I guess its usefulness is largely depending on the user space > applications. Most of them should not be sensible to it. I see, that makes some sense although it makes it advantageous to split heavy dirtier task into two less heavy dirtiers which is a bit strange. But as you say, precise results here probably do not matter much. Have a nice weekend Honza -- Jan Kara <jack@xxxxxxx> SUSE Labs, CR -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
