Re: Second regression due to libxl: Remove linux udev rules (2ba368d13893402b2f1fb3c283ddcc714659dd9b)

[Ian Campbell <ian.campbell@citrix.com>]

On Tue, 2015-08-04 at 10:14 +0200, Roger Pau Monné wrote:
> This was not a problem when using udev, because there's no timeout, but 
> libxl has a hard timeout (10s) regarding hotplug script execution. The 
> only way I see to solve this is to remove the checks done in the block 
> hotplug script, or to increase the timeout (but since the execution 
> time is not bounded this is doomed to fail if enough guests are using 
> the same image).

If we have evidence (as we appear to) that the timeout is not sufficient
for starting large numbers of guests in // then increasing the timeout
would seem reasonable.

Perhaps someone could set an insane timeout and measure the maximum time
they see in practice in this test case. Then we could decide what a
reasonable timeout should be? Maybe we could even find a linear pattern in
the number of guests and use that to extrapolate a reasonable timeout?

That would be a small self contained fix which could be made for 4.6 I
think.

For 4.7 we could consider other options. I don't think we want to remove
the timeout altogether, but perhaps the mechanisms used for doing and/or
checking the level sharing could be optimised somewhat (perhaps by reusing
the loopback device?).

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel