----- Original Message -----
> From: "Tony Asleson" <tasle...@redhat.com>
> To: email@example.com
> Sent: Thursday, April 5, 2012 6:02:31 PM
> Subject: Re: [vdsm] /etc/rc.c/init.d/vdsmd set filters for libvirt at DEBUG
> On 04/05/2012 03:42 PM, Saggi Mizrahi wrote:
> > The performance implication though existing don't really pop up in
> > normal use.
> In a previous life I worked on a product that logged a *ton* to
> circular in memory binary buffers (memory was preserved on reset and
> written to flash when the system booted to allow later retrieval aka.
> flight recorder). It was optimized to minimize run-time costs (~10us
> log a message). This was done so that you never had to re-create a
> failure. No one believed that it would effect performance that much,
> but one of the folks profiled it and found logging was eating at
> 10% of the total CPU (embedded system).
Agreed. Nothing is for free, everything has a cost.
> > It's true we usually don't need logs that far back but as I said
> > it's
> > really not an issue. Even when you try and run a 1000 VMs you will
> > not be blocked on logging trying to get written to disk I can
> > assure
> > you of that.
> Has anyone gathered the data that shows what the logging costs when
> have *many* VMs? Even a small operation can become relevant when you
> have sufficiently large numbers of them.
Working on that indirectly as we try to measure the time it takes to start up
VMs using a vdsm client.
> vdsm-devel mailing list
vdsm-devel mailing list