Re: Proposed package removals after 2016Q4

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Tue, Dec 20, 2016 at 01:55:58PM -0500, Greg Troxel wrote:
> 
> Manuel Bouyer <bouyer%antioche.eu.org@localhost> writes:
> 
> > On Tue, Dec 20, 2016 at 01:19:32PM -0500, Greg Troxel wrote:
> >> 
> >> Manuel Bouyer <bouyer%antioche.eu.org@localhost> writes:
> >> 
> >> > I'm still using 4.2 as the last version with a working xm.
> >> > I'm not sure I can switch all my servers to the xl stack yet.
> >> 
> >> Are you using old rc.d files, or is it just a matter of xend=YES and
> >> xencommons=NO?
> 
> But how are you causing xm to work?  Are you running different daemons
> than the standard recommendation?  Different rc.d files?  I would think
> you'd need the old rc.d files from before the xm/xl change.

Afaik the scripts provided with xentools42 works fine for xm.
Maybe you need to change xl to xm in xendomains but that's it.

> 
> > It's the problem that xl runs things in parallel, while xm
> > serialises them.
> 
> Do you mean that on startup, the xl-based xendomains script that is part
> of the xentools42 package causes "xl start" to be fired off for all
> domains at once, rather than how "xm start" somewhat causes the next to
> wait, perhaps by waiting until it's mostly started to return?

all the scripts to configure devices are started in parallel.
I have modified scripts (for editing /etc/ipf.conf at
domU startup, mostly), which can't be run in parallel.

We now have locking in these scripts so this should be safe, but I've
not tested with 10s of interfaces as I have on my servers.
(when I tried xl on these servers, it would time out configuring
virtual interfaces because I had too much, but the locking I added at this
time was too heavy).

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--