hosted-engine : how to shutdown hosts?
Hi, just a minor problem I guess: I have a small test environment with 2 hosts and a hosted engine on a separate NFS3 share, all running CentOS7. The VMs are running from an iSCSI storage. When I want to shutdown the environment I: - shutdown VMs - enable global maintenance mode - "shutdown -h now" on the hosted engine vm - "shutdown -h now" on the hosts The problem: instead of shutting down, the hosts perform a reboot after a short while. Is this the expected behaviour or a known bug? How can I cleanly shutdown my OVirt-environment? Thank you, Uwe
----- Original Message -----
From: "Uwe Laverenz" <uwe@laverenz.de> To: users@ovirt.org Sent: Saturday, February 28, 2015 3:41:55 PM Subject: [ovirt-users] hosted-engine : how to shutdown hosts?
Hi,
just a minor problem I guess: I have a small test environment with 2 hosts and a hosted engine on a separate NFS3 share, all running CentOS7. The VMs are running from an iSCSI storage.
When I want to shutdown the environment I:
- shutdown VMs - enable global maintenance mode - "shutdown -h now" on the hosted engine vm - "shutdown -h now" on the hosts
The problem: instead of shutting down, the hosts perform a reboot after a short while.
The hosts? Did you look at their console and see that they actually "reboot after a short while"? Or perhaps they do not shutdown at all and reboot instead?
Is this the expected behaviour or a known bug? How can I cleanly shutdown my OVirt-environment?
I think that before shutting down a host you should move it to local maintenance. Not sure, however, how it was started back. Please try and report back. Thanks! Also: Do you have fencing enabled? You wrote '"shutdown -h now" on the hosts'. What happens if you shutdown just one of them and then wait a bit? Does it get started too? Are you certain that 'shutdown -h now' worked on these hosts before the setup - that it's not some kernel/bios/etc issue? Best, -- Didi
On Sun, Mar 01, 2015 at 02:26:12AM -0500, Yedidyah Bar David wrote:
You wrote '"shutdown -h now" on the hosts'. What happens if you shutdown just one of them and then wait a bit? Does it get started too?
Are you certain that 'shutdown -h now' worked on these hosts before the setup - that it's not some kernel/bios/etc issue?
I've seen similar issues on some CentOS 7 hosts running on desktop-class hardware. I haven't got time for extensive debugging, but I'd bet it's a kernel or hardware-related issue (or a combination of both). There's nothing indicating that this is related to oVirt, at least in my case. Regards - Frank
On 02/03/15 12:25, Frank Wall wrote:
On Sun, Mar 01, 2015 at 02:26:12AM -0500, Yedidyah Bar David wrote:
You wrote '"shutdown -h now" on the hosts'. What happens if you shutdown just one of them and then wait a bit? Does it get started too?
Are you certain that 'shutdown -h now' worked on these hosts before the setup - that it's not some kernel/bios/etc issue?
I've seen similar issues on some CentOS 7 hosts running on desktop-class hardware. I haven't got time for extensive debugging, but I'd bet it's a kernel or hardware-related issue (or a combination of both). There's nothing indicating that this is related to oVirt, at least in my case.
Regards - Frank
Indeed so; there's no hardware level interaction in hosted engine. Once you put the setup in global maintenance the nodes will only monitor that state. So nothing is expected from the HA agent in case the VM or a node goes down.
On Mon, Mar 02, 2015 at 11:25:38AM +0100, Frank Wall wrote:
On Sun, Mar 01, 2015 at 02:26:12AM -0500, Yedidyah Bar David wrote:
You wrote '"shutdown -h now" on the hosts'. What happens if you shutdown just one of them and then wait a bit? Does it get started too?
Are you certain that 'shutdown -h now' worked on these hosts before the setup - that it's not some kernel/bios/etc issue?
I've seen similar issues on some CentOS 7 hosts running on desktop-class hardware. I haven't got time for extensive debugging, but I'd bet it's a kernel or hardware-related issue (or a combination of both). There's nothing indicating that this is related to oVirt, at least in my case.
FWIW, after upgrading to CentOS 7.1.1503 (and oVirt 3.5.3) this issue is gone, at least for me. Regards - Frank
----- Original Message -----
From: "Frank Wall" <fw@moov.de> To: users@ovirt.org Sent: Monday, June 22, 2015 11:41:02 AM Subject: Re: [ovirt-users] hosted-engine : how to shutdown hosts?
On Mon, Mar 02, 2015 at 11:25:38AM +0100, Frank Wall wrote:
On Sun, Mar 01, 2015 at 02:26:12AM -0500, Yedidyah Bar David wrote:
You wrote '"shutdown -h now" on the hosts'. What happens if you shutdown just one of them and then wait a bit? Does it get started too?
Are you certain that 'shutdown -h now' worked on these hosts before the setup - that it's not some kernel/bios/etc issue?
I've seen similar issues on some CentOS 7 hosts running on desktop-class hardware. I haven't got time for extensive debugging, but I'd bet it's a kernel or hardware-related issue (or a combination of both). There's nothing indicating that this is related to oVirt, at least in my case.
FWIW, after upgrading to CentOS 7.1.1503 (and oVirt 3.5.3) this issue is gone, at least for me.
Thanks for the update :-) Best, -- Didi
I think it's a systemd related issue: http://cgit.freedesktop.org/systemd/systemd/commit/?id=4b5d8d0f22ae61ceb45a2... I got that behaviour in a completely different environment (gentoo server, totally unrelated to ovirt), and it seems that systemd-halt.service is tweaking in a bad manner with command line parameters. You can try to install a newer systemd version. Hope this helps, gc Il 28/02/2015 14:41, Uwe Laverenz ha scritto:
Hi,
just a minor problem I guess: I have a small test environment with 2 hosts and a hosted engine on a separate NFS3 share, all running CentOS7. The VMs are running from an iSCSI storage.
When I want to shutdown the environment I:
- shutdown VMs - enable global maintenance mode - "shutdown -h now" on the hosted engine vm - "shutdown -h now" on the hosts
The problem: instead of shutting down, the hosts perform a reboot after a short while.
Is this the expected behaviour or a known bug? How can I cleanly shutdown my OVirt-environment?
Thank you, Uwe _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
-- Giulio Casella giulio at di.unimi.it System and network manager Computer Science Dept. - University of Milano
participants (5)
-
Doron Fediuck -
Frank Wall -
Giulio Casella -
Uwe Laverenz -
Yedidyah Bar David