Date: Tue, 10 Nov 2015 10:34:58 +0100 From: Michael Reifenberger <mike@reifenberger.com> To: Roger Pau =?utf-8?b?TW9ubsOp?= <roger.pau@citrix.com> Cc: xen@freebsd.org Subject: Re: Current shortcomings (XEN dom0 & FreeBSD) Message-ID: <20151110103458.Horde.H363qPg0Pt7dxArnT4Xwun9@mail.eeeit.de> In-Reply-To: <563CBFA7.30502@citrix.com> References: <20150918194154.Horde.PQcchwucJFPQY4U0K75MgpW@mail.eeeit.de> <56027D61.70207@citrix.com> <20150923163642.Horde.C2gq8tfwkC45mOK8NwCIAj-@mail.eeeit.de> <5602BD02.7050004@citrix.com> <20151106132308.Horde.wqlInItwP9c68Ih_23aoBbQ@mail.eeeit.de> <563CBFA7.30502@citrix.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Zitat von Roger Pau Monné <roger.pau@citrix.com>:
> Hello,
>
> El 06/11/15 a les 13.23, Michael Reifenberger ha escrit:
>> Hi,
>> before analyzing deeper I just wanted to ask if the following symptoms
>> are already known:
>>
>> - Often / Mostly the FreeBSD dom0 hangs without further output during
>> poweroff (reboot works though)
>
> Yes, I'm aware of this issue, unfortunately the way to solve it it's not
> clear.
>
> In order to do a proper power off when running as Dom0 we would have to
> modify ACPICA code, which is a separate upstream project that's used by
> a bunch of different OSes. Linux does it this way, but then they have to
> keep all this local modifications on top of upstream ACPICA, which
> increases the maintainership burden.
>
How big are the diffs to the original version?
Have you asked upstream already if the Patches would be accepted?
>> - The same happens mostly for the Centos7 guests I'm currently testing.
>
> Do you mean that CentOS 7 guests do not poweroff properly? Are those PV
> or HVM guests?
>
Yes, as it seems. They are HVM for first installation and PV afterwards.
BTW: How can one specify pv disks to a HVM guest (Some guests get more
than 4 disks)?
Something like the following is not accepted:
...
disk= [
'/dev/zvol/zdata/VM/cnt_nw1/root,raw,hda,w',
'/dev/zvol/zdata/VM/cnt_nw1/swap,raw,xvdb,w',
'/dev/zvol/zdata/VM/cnt_nw1/sapmnt,raw,xvdc,w',
'/VM/ISO/CentOS-7-x86_64-Minimal-1503-01.iso,raw,hdc:cdrom,r'
...
> I usually use Debian guests when I have to test Linux and have never
> experienced this.
>
SUSE SLES11SP2 also seems not affected.
>> - Sometimes (usually after some uptime and/or many guest reboots) the
>> blockback
>> (ZFS Zvols in my case) seems to get wedged (guest hang during startup)...
>
> Yes, I've also experienced this once or twice. The problem is that the
> blkback error path can get deadlocked depending on the situation. I have
> to look into cleaning it when I have some time.
>
Since bapt@ stepped down from maintaining the xen ports:
Has someone already tried to use xen 4.6 kernel and tools?
Thanks!
Greetings
---
Michael
Gruß
---
Michael Reifenberger
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20151110103458.Horde.H363qPg0Pt7dxArnT4Xwun9>