Re: [Users] where do bond interfaces come from when adding a node that is FC17
On 09/20/2012 08:10 AM, Mark Wu wrote: On 09/19/2012 11:25 PM, Christopher Maestas wrote: When you add a fc17 node and look at it, it seems to add: * ovirtmgmt bridge * p2p1 interface * and bond0-bond3. Where does the bonding information get stored? It doesn't seem to be in /etc/sysconfig/network-scripts? The bonding devices are created dynamically when vdsmd starts. The reason I ask, is I have two nodes one of which has the bonding interfaces up and the other which doesn't. The one that doesn't have the bonding interfaces up is failing to be added to the ovirt node list. I am not sure how the failure of adding host is related to bonding interfaces. You could check the log /tmp/vds_bootstrap.xx.log and /tmp/vds_installer.xx.log during node installation. Why doesn't vdsm write its log files to /var/log/vdsm where people would expect them to be? I think writing this to /tmp is a really bad habit that should be fixed sooner rather than later especially given the importance of the logs when something goes wrong. Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] ovirt vs openstack
On 09/07/2012 08:46 PM, Jason Brooks wrote: On 09/07/2012 08:59 AM, Neo Vlanik wrote: Hi, Forgive me if this is a dump question. What is the difference between ovirt and openstack projects? I think of it like this: OpenStack is / aims to be an open source Amazon Web Services (EC2, S3, etc.), oVirt is / aims to be an open source vSphere. The former tends toward more automation, the latter toward more flexibility. There's a good deal of overlap though. That's my take, I'm interested to hear others' viewpoints. Also, I've seen Kyle Mestery blog on the topic: http://www.siliconloons.com/?p=179. I think both technologies are very complementary. OpenStack currently currently has no good management interface and I look at oVirt as mostly a cloud management interface. I can see a future in which oVirt is used to manage a bunch of openstack nodes. Quantum integration is already happening and User management is already external to oVirt so keystone can probably plugged in there at some point. Then either vdsm needs to learn to talk to nova or oVirt needs to learn to talk to nova as an alternative to vdsm. Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] sanlock issues
On 08/07/2012 04:08 PM, Jacob Wyatt wrote: Solved with a hack. oVirt Node Hypervisor release 2.5.0 (2.0.fc17) Couldn't start a VM because sanlock wasn't running. Sanlock wasn't running because wdmd wasn't running. wdmd was running because softdog kernel module wasn't loaded. As I didn't know of another way of making the change persistent I edited /config/usr/sbin/ifup and added /sbin/modprobe softdog to the top of the script. I really think it would be a good idea for the oVirt Node team to add in some persistent script files that are run at various points in the boot process so that people like me can add in hacks where needed to make it work. Shouldn't the following issue have fixed this already? https://bugzilla.redhat.com/show_bug.cgi?id=832935 Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] Node as guest - Nested KVM
On 08/02/2012 03:25 PM, Mike Burns wrote: On Thu, 2012-08-02 at 09:32 -0400, Douglas Landgraf wrote: Hi, I remember someone asking about this topic few weeks ago, here the steps: http://wiki.ovirt.org/wiki/Vdsm_Developers#Running_Node_as_guest_-_Nested_KVM Also check out a slightly more generic: http://kashyapc.wordpress.com/2012/01/14/nested-virtualization-with-kvm-intel/ BTW, with intel, much simpler to add a file in /etc/modprobe.d : $ cat /etc/modprobe.d/kvm.conf options kvm_intel nested=Y Mike But does this actually work? The last time I tried (a few weeks ago) I wasn't able to install a F17 guest inside a F17 guest. The whole thing just hung early in the installation. Similar results have been reported here (a follow up post to the one you posted above): http://kashyapc.wordpress.com/2012/07/28/nested-virtualization-with-intel-take-2-with-fedora-17/ Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] Sanlock issue when trying to start vm
On 06/21/2012 12:04 AM, Robert Middleswarth wrote: On 06/20/2012 08:33 AM, Dennis Jacobfeuerborn wrote: On 06/20/2012 08:17 AM, Rami Vaknin wrote: On 06/19/2012 09:36 PM, Dennis Jacobfeuerborn wrote: Hi, after getting the 3.1 beta engine and a host set up I now get an error when trying to start a vm. Engine reports this: VM myvm is down. Exit message: internal error Failed to open socket to sanlock daemon: No such file or directory. Please take a look at the following bugs: https://bugzilla.redhat.com/show_bug.cgi?id=832935 https://bugzilla.redhat.com/show_bug.cgi?id=832056 Thanks, after modprobe softdog and restarting wdmd and sanlock I was able to start the VM. Unfortunately my experiment with using nested VMs to fake nodes didn't pan out (the guest-in-guest booted for a bit and then froze). Is there a way to use pure qemu guests for testing (like in devstack)? While I do have two systems that support hardware virtualization i cannot reinstall either of them to use them as a host. Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users The CentOS builds include a plugin called vdsm-hook-simpleqemu I think I saw the hook sitting in Git as well so that plug-in might do what you are looking for? Hm, looks like it but that doesn't seem to be available for the regular builds. The is a vdsm-hook-faqemu package though which when comparing the code seems to do a similar thing (it also uses the fake_kvm_support setting from vdsm.conf like simpleqemu). Is the latter maybe a replacement of the former? Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] Sanlock issue when trying to start vm
On 06/20/2012 08:17 AM, Rami Vaknin wrote: On 06/19/2012 09:36 PM, Dennis Jacobfeuerborn wrote: Hi, after getting the 3.1 beta engine and a host set up I now get an error when trying to start a vm. Engine reports this: VM myvm is down. Exit message: internal error Failed to open socket to sanlock daemon: No such file or directory. Please take a look at the following bugs: https://bugzilla.redhat.com/show_bug.cgi?id=832935 https://bugzilla.redhat.com/show_bug.cgi?id=832056 Thanks, after modprobe softdog and restarting wdmd and sanlock I was able to start the VM. Unfortunately my experiment with using nested VMs to fake nodes didn't pan out (the guest-in-guest booted for a bit and then froze). Is there a way to use pure qemu guests for testing (like in devstack)? While I do have two systems that support hardware virtualization i cannot reinstall either of them to use them as a host. Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
[Users] Guest life-cycle
Hi, after being able to start my first guest I noticed two things: 1) When I started the guest with the run once option the console button didn't become available. Navigating to another section of the GUI and back didn't help. I was only able to make is available by closing the browser tab and then going to the admin portal again from scratch. 2) When my guest froze (due to nested virtualization experimentation that apparently doesn't work yet) I had no (obvious) way to kill the vm. Hitting the stop button apparently only sends a shutdown command but I couldn't find any way to issue a destroy command. After killing the qemu process on the host manually the guest in the portal gui apparently went into an unknown ? state which isn't surprising but now there seems to be no way to get rid of the vm. What is proper way to get rid of such a misbehaving guest? Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] Guest life-cycle
On 06/20/2012 02:45 PM, Andrew Cathrow wrote: - Original Message - From: Dennis Jacobfeuerborn denni...@conversis.de To: users users@ovirt.org Sent: Wednesday, June 20, 2012 8:40:29 AM Subject: [Users] Guest life-cycle Hi, after being able to start my first guest I noticed two things: 1) When I started the guest with the run once option the console button didn't become available. Navigating to another section of the GUI and back didn't help. I was only able to make is available by closing the browser tab and then going to the admin portal again from scratch. there was an issue with the spice-xpi that was fixed in F17, I'd have to check to see if that made it to F16. What OS are you on and what version of spice-xpi ? The client is on F15 with spice-xpi-2.7-1. Should this affect the icon in the gui though? I thought spice would only come into play once I actually click that button and start the console. 2) When my guest froze (due to nested virtualization experimentation that apparently doesn't work yet) I had no (obvious) way to kill the vm. Hitting the stop button apparently only sends a shutdown command but I couldn't find any way to issue a destroy command. Shutdown requires guest agent. If you right click on the VM you can do a stop - which forces it down. This only seems to be available as long as the VM is still running. Now that it is in an unknown state that menu option is grayed out as is the remove option. How can I clean this up? Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
[Users] Sanlock issue when trying to start vm
Hi, after getting the 3.1 beta engine and a host set up I now get an error when trying to start a vm. Engine reports this: VM myvm is down. Exit message: internal error Failed to open socket to sanlock daemon: No such file or directory. On the node no sanlock daemon is running. When I try to start the service I get this: Jun 19 20:22:31 node systemd-sanlock[13607]: Starting sanlock: [ OK ] Jun 19 20:22:31 node sanlock[13621]: 2012-06-19 20:22:31+0200 910 [13621]: sanlock daemon started 2.3 aio 1 10 renew 20 80 host 93bea910-1d9d-4203-8333-beb3d7b92c10.node.local time 1340130151 Jun 19 20:22:31 node sanlock[13621]: 2012-06-19 20:22:31+0200 910 [13621]: set scheduler RR|RESET_ON_FORK priority 99 failed: Operation not permitted Jun 19 20:22:31 node sanlock[13621]: 2012-06-19 20:22:31+0200 910 [13621]: wdmd connect failed for watchdog handling Jun 19 20:22:31 node systemd[1]: sanlock.service: main process exited, code=exited, status=255 Jun 19 20:22:31 node systemd[1]: Unit sanlock.service entered failed state. I followed the SELinux and sanlock threads but I disabled SELinux on the node so that cannot be a reason for this to fail. Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
[Users] Host setup for 3.1 beta
Hi, is there a working way to add a host with the current beta code? I tried the latest node iso and also a plain Fedora 17 with the latest vdsm packages but in both cases the registration seems to succeed but the installation seems to fail. Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] Host setup for 3.1 beta
On 06/18/2012 10:27 PM, Dan Kenigsberg wrote: On Mon, Jun 18, 2012 at 10:24:51PM +0200, Dennis Jacobfeuerborn wrote: Hi, is there a working way to add a host with the current beta code? I tried the latest node iso and also a plain Fedora 17 with the latest vdsm packages but in both cases the registration seems to succeed but the installation seems to fail. Could you provide more information about the failure? is this the issue reported by Trey Dockendorf in http://lists.ovirt.org/pipermail/users/2012-June/002466.html Actually after seeing the /tmp/vds_bootstrap.64713.log bit in that mail I was able to find out what the problem was and fix it. Logs should really go into /var/log that where I was trying to find information on what was going on. The installation seems to succeed now and the Host is displayed as Up. I got stuck almost 5 minutes in the state Reboot though even though the node was up in a few seconds after the reboot. What exactly is causing this long delay? Is the engine trying to contact vdsm or is the engine waiting for vdsm to connect to the engine? Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] Host setup for 3.1 beta
On 06/19/2012 01:00 AM, Robert Middleswarth wrote: Hard coded time out. The system waits five min before tiring. Thanks Robert On 06/18/2012 05:58 PM, Dennis Jacobfeuerborn wrote: On 06/18/2012 10:27 PM, Dan Kenigsberg wrote: On Mon, Jun 18, 2012 at 10:24:51PM +0200, Dennis Jacobfeuerborn wrote: Hi, is there a working way to add a host with the current beta code? I tried the latest node iso and also a plain Fedora 17 with the latest vdsm packages but in both cases the registration seems to succeed but the installation seems to fail. Could you provide more information about the failure? is this the issue reported by Trey Dockendorf in http://lists.ovirt.org/pipermail/users/2012-June/002466.html Actually after seeing the /tmp/vds_bootstrap.64713.log bit in that mail I was able to find out what the problem was and fix it. Logs should really go into /var/log that where I was trying to find information on what was going on. The installation seems to succeed now and the Host is displayed as Up. I got stuck almost 5 minutes in the state Reboot though even though the node was up in a few seconds after the reboot. What exactly is causing this long delay? Is the engine trying to contact vdsm or is the engine waiting for vdsm to connect to the engine? Yes I figured as much but why does it time out? Who is responsible for re-establishing the connection after the reboot and why is that timing out? Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] oVirt 3.1 Release Test Day
On 06/13/2012 11:36 PM, Mike Burns wrote: It's been pointed out to me that there has been a lack of communication around the Test Day for the next oVirt Release. I apologize for the oversight that information hasn't been posted to these lists previously. When is the test day? Thursday 2012-06-14 Where can you find info on the test day? http://ovirt.org/wiki/Testing/OvirtTestDay3.1 The link for the pre-built node image points to http://ovirt.org/releases/nightly/binary/ovirt-node-image-2.2.3-1.1.fc16.iso Is that really the correct image to test? Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
[Users] cannot install node iso images
Hi, I just tried installing the 2.3.0 node iso images from http://ovirt.org/releases/nightly/binary/ in a VM on a Fedora 17 Host and failed. First the image takes 1-2 minutes to boot and I'm not sure why it would take so long. There is not output made so it's hard to say what's going on even when you press esc the skip the plymouth progress bar. Second after that I get quite a few errors about denied permissions and read-only filesystems. For example systemd tries to create /var/log/wtmp but fails with permission denied and it tries to remove /var/lib/rpm/__db.00* files which gets a read-only filesystem error. Third once it actually gets to the Install hypervisor TUI this interface does not respond to key presses. When I press the cursor keys I only get text like ^[[C plastered on the screen. Actually starting the installation is impossible. What am I doing wrong? Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] Steal Time Implemented?
On 05/29/2012 04:33 AM, Andrew Cathrow wrote: - Original Message - From: Schorschi schors...@dc.rr.com To: users@ovirt.org Sent: Sunday, May 27, 2012 10:15:05 PM Subject: [Users] Steal Time Implemented? Given 'Steal Time' was introduced in RHEL 6.3? How long until same feature set will be in oVirt Nodes? Or is it already integrated? Would we implement it in the node, or would this be something that we'd report in the guest agent and pass back through vdsm to the engine. Steal would be reported by the guests. For the nodes you can report guest though I'm not sure if this is part of RHEL yet. This is what it looks like on a Fedora 15 system (man mpstat): %steal Show the percentage of time spent in involuntary wait by the virtual CPU or CPUs while the hypervisor was servicing another virtual processor. %guest Show the percentage of time spent by the CPU or CPUs to run a virtual processor. Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] Password reset
On 05/14/2012 07:20 AM, Oved Ourfalli wrote: You can use the engine-config utlilty. It is used to show/set config entries. Currently, the admin password is a config entry, so you can do the following: engine-config -s AdminPassword=your_password and it will do the trick. That worked as expected, thanks! Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
[Users] Password reset
Hi, how can I reset the password of the admin@internal user? Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] Adding users
On 05/08/2012 01:44 PM, Yair Zaslavsky wrote: On 05/08/2012 02:26 PM, Dennis Jacobfeuerborn wrote: Hi, I just installed oVirt and so far things seem to be running smoothly for the 1-node setup. I can't seem to add users though. When I go to Users-add I only get a search dialog that lists the admin user no matter what I enter as search string and there seems to be no obvious way to actually add a new user. Clicking ok just dismisses the dialog. What user are you using to log in? Have you used engine-manage-domains in order to add an LDAP/Kerberos domain? I used the user admin that is the default with the regular rpm install. The UI doesn't mention anything about a LDAP/Kerberos domain when I try to add the user. Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
[Users] Adding users
Hi, I just installed oVirt and so far things seem to be running smoothly for the 1-node setup. I can't seem to add users though. When I go to Users-add I only get a search dialog that lists the admin user no matter what I enter as search string and there seems to be no obvious way to actually add a new user. Clicking ok just dismisses the dialog. Also after installing my first VM, stopping it and starting it up again the console icon stayed greyed out. I actually had to restart my browser and log back in in order to make the console icon clickable again. Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Re: [Users] error when regestering a host with vdsm 4.9.6 in ovirt 3.1
On 05/03/2012 08:34 PM, Itamar Heim wrote: On 05/03/2012 08:43 PM, Dennis Jacobfeuerborn wrote: On 05/03/2012 04:36 PM, Doron Fediuck wrote: Hi Max, MainThread::DEBUG::2012-05-03 11:22:33,866::vdsm-reg-setup::73::root::validate end. return: False means your configuration there is invalid. That message should be changed so it becomes clear that there is a problem with the configuration. Speaking as an admin these kind of cryptic log entries are really frustrating to deal with. please open a bugzilla and suggest a more descriptive error. (as always, even better if you also send the patch to fix it to gerrit) Filed here: https://bugzilla.redhat.com/show_bug.cgi?id=818712 Regards, Dennis ___ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users