Re: [SPDK] Problem with Blobstore when write 65MB continously

[Walker, Benjamin <benjamin.walker at intel.com>]

[-- Attachment #1: Type: text/plain, Size: 3822 bytes --]

On Wed, 2018-01-10 at 17:02 +0000, Luse, Paul E wrote:
> OK yeah that makes sense, one could experiment with batching some number of
> requests and counting through callbacks to find a good number to submit before
> returning, was hoping for something like spdk_yield_thread() but of course
> didn't look first.  Do you think that might be useful or has this never come
> up before? 

If we could implement spdk_yield_thread() in C, it would be amazing.
Unfortunately, coroutines are impossible to implement in pure C. It makes a lot
of sense to program all of this asynchronous stuff using coroutines, or even
better futures and promises, but those things just aren't available to us. And
sticking to pure, standard compliant C is important for SPDK because it gets
ported to all sorts of architectures and run time environments.

> 
> Thx
> Paul
> 
> -----Original Message-----
> From: SPDK [mailto:spdk-bounces(a)lists.01.org] On Behalf Of Walker, Benjamin
> Sent: Wednesday, January 10, 2018 9:58 AM
> To: freeman.zhang1992(a)gmail.com; spdk(a)lists.01.org
> Subject: Re: [SPDK] Problem with Blobstore when write 65MB continously
> 
> On Wed, 2018-01-10 at 16:47 +0000, Luse, Paul E wrote:
> > Damn, I should have known that :( Thanks guys!! 
> > 
> > So I'll take the easy way out and just ask.. what's the most efficient 
> > way for an app, designed just this one for whatever reason, 
> > essentially relinquish control to the event reactor - is there a 
> > single call he can slip into the loop to give other pending events, if 
> > any a chance to run and if not continue on submitting?
> 
> Return from the function submitting I/O after submitting a certain number and
> resume submitting new I/O in response to completion callbacks.
> 
> > 
> > Thx
> > Paul
> > 
> > -----Original Message-----
> > From: SPDK [mailto:spdk-bounces(a)lists.01.org] On Behalf Of Walker, 
> > Benjamin
> > Sent: Wednesday, January 10, 2018 9:33 AM
> > To: freeman.zhang1992(a)gmail.com; spdk(a)lists.01.org
> > Subject: Re: [SPDK] Problem with Blobstore when write 65MB continously
> > 
> > On Wed, 2018-01-10 at 16:21 +0000, Harris, James R wrote:
> > > Hi Paul and Zhengyu,
> > > 
> > > The problem is that the app is not giving the block device a chance 
> > > to complete any I/O while submitting the 520 back-to-back requests.
> > > Blobstore is passive here – it does not do any polling on the block 
> > > device – that is up to the application.
> > 
> > To additionally clarify - the bdev layer will poll for completions on 
> > your behalf, but it does so on the same thread that you are submitting 
> > I/O from. If you are in a tight loop submitting I/O and never yield 
> > back to the event reactor, the polling won't have a chance to occur. 
> > You can either increase the number of reqs per channel or submit 
> > smaller batches at a time. Basically, do what Jim said below.
> > 
> > > Increasing the number of channel reqs would work – but at some point 
> > > these will still run out.  So it really depends on your application 
> > > – either increase the channel reqs to the absolutely maximum you 
> > > will ever need, or add ENOMEM handling.
> > 
> > _______________________________________________
> > SPDK mailing list
> > SPDK(a)lists.01.org
> > https://lists.01.org/mailman/listinfo/spdk
> > _______________________________________________
> > SPDK mailing list
> > SPDK(a)lists.01.org
> > https://lists.01.org/mailman/listinfo/spdk
> 
> _______________________________________________
> SPDK mailing list
> SPDK(a)lists.01.org
> https://lists.01.org/mailman/listinfo/spdk
> _______________________________________________
> SPDK mailing list
> SPDK(a)lists.01.org
> https://lists.01.org/mailman/listinfo/spdk