Re: Balloon pressuring page cache

["Michael S. Tsirkin" <mst@redhat.com>]

On Thu, Jan 30, 2020 at 11:59:46AM -0800, Tyler Sanderson wrote:
> 
> 
> On Thu, Jan 30, 2020 at 7:31 AM Wang, Wei W <wei.w.wang@intel.com> wrote:
> 
>     On Thursday, January 30, 2020 11:03 PM, David Hildenbrand wrote:
>     > On 29.01.20 20:11, Tyler Sanderson wrote:
>     > >
>     > >
>     > > On Wed, Jan 29, 2020 at 2:31 AM David Hildenbrand <david@redhat.com
>     > > <mailto:david@redhat.com>> wrote:
>     > >
>     > >     On 29.01.20 01:22, Tyler Sanderson via Virtualization wrote:
>     > >     > A primary advantage of virtio balloon over other memory reclaim
>     > >     > mechanisms is that it can pressure the guest's page cache into
>     > >     shrinking.
>     > >     >
>     > >     > However, since the balloon driver changed to using the shrinker
>     API
>     > >     >
>     > >
>     > <https://github.com/torvalds/linux/commit/71994620bb25a8b109388fefa9
>     > e99a28e355255a#diff-fd202acf694d9eba19c8c64da3e480c9> this
>     > >     > use case has become a bit more tricky. I'm wondering what the
>     > intended
>     > >     > device implementation is.
>     > >     >
>     > >     > When inflating the balloon against page cache (i.e. no free
>     memory
>     > >     > remains) vmscan.c will both shrink page cache, but also invoke
>     the
>     > >     > shrinkers -- including the balloon's shrinker. So the balloon
>     driver
>     > >     > allocates memory which requires reclaim, vmscan gets this memory
>     > by
>     > >     > shrinking the balloon, and then the driver adds the memory back
>     to
>     > the
>     > >     > balloon. Basically a busy no-op.
> 
>     Per my understanding, the balloon allocation won’t invoke shrinker as
>     __GFP_DIRECT_RECLAIM isn't set, no?
> 
> I could be wrong about the mechanism, but the device sees lots of activity on
> the deflate queue. The balloon is being shrunk. And this only starts once all
> free memory is depleted and we're inflating into page cache.

So given this looks like a regression, maybe we should revert the
patch in question 71994620bb25 ("virtio_balloon: replace oom notifier with shrinker")
Besides, with VIRTIO_BALLOON_F_FREE_PAGE_HINT
shrinker also ignores VIRTIO_BALLOON_F_MUST_TELL_HOST which isn't nice
at all.

So it looks like all this rework introduced more issues than it
addressed ...

I also CC Alex Duyck for an opinion on this.
Alex, what do you use to put pressure on page cache?


> 
> 
>     > >     >
>     > >     > If file IO is ongoing during this balloon inflation then the page
>     > >     cache
>     > >     > could be growing which further puts "back pressure" on the
>     balloon
>     > >     > trying to inflate. In testing I've seen periods of > 45 seconds
>     where
>     > >     > balloon inflation makes no net forward progress.
> 
>     I think this is intentional (but could be improved). As inflation does not
>     stop when the allocation fails (it simply sleeps for a while and resumes..
>     repeat till there are memory to inflate)
>     That's why you see no inflation progress for long time under memory
>     pressure.
> 
> As noted above the deflate queue is active, so it's not just memory allocation
> failures.
>  
> 
> 
> 
>     Best,
>     Wei
>