On Tue, May 23, 2017 at 10:12:06 +0200, Martin Kletzander wrote:
On Mon, May 22, 2017 at 06:00:13PM +0200, Peter Krempa wrote:
> Hint that the users should limit the number of VMs queried in the bulk
> stats API.
> ---
> src/libvirt-domain.c | 7 +++++++
> 1 file changed, 7 insertions(+)
[...]
Is there a discussion about how to continue for the future? I
remember
Michal starting some discussion, but I'm not sure whether that was about
the same thing. Anyway, that's just a rhetorical question so that I can
say the two ideas I have:
1) Just increase the limit over time. Computers and networks are
getting faster, there's more storage space and memory, and so on.
It only makes sense to do scale other things respectively.
2) Have a new API that streams the data back over virStream. We can
then do it continuously (like every X seconds), that might help
management apps as well because I suspect that's how they use this
API anyway.
3) danpb also pointed out that we could use a chunk of shared memory to
transport the stats for local connections, so that we skip the RPC layer
altoghether and also avoid having two copies of the same data.
But that's something we need to design first.