Hi Martin, Thanks for the reply. I have done this, and the deployment completed without error. However, it still will not allow the Hosted Engine migrate to another host. The /etc/ovirt-hosted-engine/hosted-engine.conf got created ok on the host I re-installed, but the ovirt-ha-broker.service, though it starts, reports: --------------------8<------------------- Jun 27 14:58:26 kvm-ldn-01 systemd[1]: Starting oVirt Hosted Engine High Availability Communications Broker... Jun 27 14:58:27 kvm-ldn-01 ovirt-ha-broker[6101]: ovirt-ha-broker ovirt_hosted_engine_ha.broker.storage_broker.StorageBroker ERROR Failed to read metadata from /rhev/data-center/mnt/blockSD/207221b2-959b-426b-b945-18e1adfed62f/ha_agent/hosted-engine.metadata Traceback (most recent call last): File "/usr/lib/python2.7/site-packages/ovirt_hosted_engine_ha/broker/storage_broker.py", line 129, in get_raw_stats_for_service_type f = os.open(path, direct_flag | os.O_RDONLY | os.O_SYNC) OSError: [Errno 2] No such file or directory: '/rhev/data-center/mnt/blockSD/207221b2-959b-426b-b945-18e1adfed62f/ha_agent/hosted-engine.metadata' --------------------8<------------------- I checked the path, and it exists. I can run 'less -f' on it fine. The perms are slightly different on the host that is running the VM vs the one that is reporting errors (600 vs 660), ownership is vdsm:qemu. Is this a san locking issue? Thanks for any help, Cam On Tue, Jun 27, 2017 at 1:41 PM, Martin Sivak <msivak@redhat.com> wrote:
Should it be? It was not in the instructions for the migration from bare-metal to Hosted VM
The hosted engine will only migrate to hosts that have the services running. Please put one other host to maintenance and select Hosted engine action: DEPLOY in the reinstall dialog.
Best regards
Martin Sivak
On Tue, Jun 27, 2017 at 1:23 PM, cmc <iucounu@gmail.com> wrote:
I changed the 'os.other.devices.display.protocols.value.3.6 = spice/qxl,vnc/cirrus,vnc/qxl' line to have the same display protocols as 4 and the hosted engine now appears in the list of VMs. I am guessing the compatibility version was causing it to use the 3.6 version. However, I am still unable to migrate the engine VM to another host. When I try putting the host it is currently on into maintenance, it reports:
Error while executing action: Cannot switch the Host(s) to Maintenance mode. There are no available hosts capable of running the engine VM.
Running 'hosted-engine --vm-status' still shows 'Engine status: unknown stale-data'.
The ovirt-ha-broker service is only running on one host. It was set to 'disabled' in systemd. It won't start as there is no /etc/ovirt-hosted-engine/hosted-engine.conf on the other two hosts. Should it be? It was not in the instructions for the migration from bare-metal to Hosted VM
Thanks,
Cam
On Thu, Jun 22, 2017 at 1:07 PM, cmc <iucounu@gmail.com> wrote:
Hi Tomas,
So in my /usr/share/ovirt-engine/conf/osinfo-defaults.properties on my engine VM, I have:
os.other.devices.display.protocols.value = spice/qxl,vnc/vga,vnc/qxl,vnc/cirrus os.other.devices.display.protocols.value.3.6 = spice/qxl,vnc/cirrus,vnc/qxl
That seems to match - I assume since this is 4.1, the 3.6 should not apply
Is there somewhere else I should be looking?
Thanks,
Cam
On Thu, Jun 22, 2017 at 11:40 AM, Tomas Jelinek <tjelinek@redhat.com> wrote:
On Thu, Jun 22, 2017 at 12:38 PM, Michal Skrivanek <michal.skrivanek@redhat.com> wrote:
On 22 Jun 2017, at 12:31, Martin Sivak <msivak@redhat.com> wrote:
Tomas, what fields are needed in a VM to pass the check that causes the following error?
>>>> WARN [org.ovirt.engine.core.bll.exportimport.ImportVmCommand] >>>> (org.ovirt.thread.pool-6-thread-23) [] Validation of action >>>> 'ImportVm' >>>> failed for user SYSTEM. Reasons: VAR__ACTION__IMPORT >>>> >>>> ,VAR__TYPE__VM,ACTION_TYPE_FAILED_ILLEGAL_VM_DISPLAY_TYPE_IS_NOT_SUPPORTED_BY_OS
to match the OS and VM Display type;-) Configuration is in osinfo….e.g. if that is import from older releases on Linux this is typically caused by the cahgen of cirrus to vga for non-SPICE VMs
yep, the default supported combinations for 4.0+ is this: os.other.devices.display.protocols.value = spice/qxl,vnc/vga,vnc/qxl,vnc/cirrus
Thanks.
On Thu, Jun 22, 2017 at 12:19 PM, cmc <iucounu@gmail.com> wrote: > Hi Martin, > >> >> just as a random comment, do you still have the database backup from >> the bare metal -> VM attempt? It might be possible to just try again >> using it. Or in the worst case.. update the offending value there >> before restoring it to the new engine instance. > > I still have the backup. I'd rather do the latter, as re-running the > HE deployment is quite lengthy and involved (I have to re-initialise > the FC storage each time). Do you know what the offending value(s) > would be? Would it be in the Postgres DB or in a config file > somewhere? > > Cheers, > > Cam > >> Regards >> >> Martin Sivak >> >> On Thu, Jun 22, 2017 at 11:39 AM, cmc <iucounu@gmail.com> wrote: >>> Hi Yanir, >>> >>> Thanks for the reply. >>> >>>> First of all, maybe a chain reaction of : >>>> WARN [org.ovirt.engine.core.bll.exportimport.ImportVmCommand] >>>> (org.ovirt.thread.pool-6-thread-23) [] Validation of action >>>> 'ImportVm' >>>> failed for user SYSTEM. Reasons: VAR__ACTION__IMPORT >>>> >>>> ,VAR__TYPE__VM,ACTION_TYPE_FAILED_ILLEGAL_VM_DISPLAY_TYPE_IS_NOT_SUPPORTED_BY_OS >>>> is causing the hosted engine vm not to be set up correctly and >>>> further >>>> actions were made when the hosted engine vm wasnt in a stable state. >>>> >>>> As for now, are you trying to revert back to a previous/initial >>>> state ? >>> >>> I'm not trying to revert it to a previous state for now. This was a >>> migration from a bare metal engine, and it didn't report any error >>> during the migration. I'd had some problems on my first attempts at >>> this migration, whereby it never completed (due to a proxy issue) but >>> I managed to resolve this. Do you know of a way to get the Hosted >>> Engine VM into a stable state, without rebuilding the entire cluster >>> from scratch (since I have a lot of VMs on it)? >>> >>> Thanks for any help. >>> >>> Regards, >>> >>> Cam >>> >>>> Regards, >>>> Yanir >>>> >>>> On Wed, Jun 21, 2017 at 4:32 PM, cmc <iucounu@gmail.com> wrote: >>>>> >>>>> Hi Jenny/Martin, >>>>> >>>>> Any idea what I can do here? The hosted engine VM has no log on any >>>>> host in /var/log/libvirt/qemu, and I fear that if I need to put the >>>>> host into maintenance, e.g., to upgrade it that I created it on >>>>> (which >>>>> I think is hosting it), or if it fails for any reason, it won't get >>>>> migrated to another host, and I will not be able to manage the >>>>> cluster. It seems to be a very dangerous position to be in. >>>>> >>>>> Thanks, >>>>> >>>>> Cam >>>>> >>>>> On Wed, Jun 21, 2017 at 11:48 AM, cmc <iucounu@gmail.com> wrote: >>>>>> Thanks Martin. The hosts are all part of the same cluster. >>>>>> >>>>>> I get these errors in the engine.log on the engine: >>>>>> >>>>>> 2017-06-19 03:28:05,030Z WARN >>>>>> [org.ovirt.engine.core.bll.exportimport.ImportVmCommand] >>>>>> (org.ovirt.thread.pool-6-thread-23) [] Validation of action >>>>>> 'ImportVm' >>>>>> failed for user SYST >>>>>> EM. Reasons: >>>>>> >>>>>> VAR__ACTION__IMPORT,VAR__TYPE__VM,ACTION_TYPE_FAILED_ILLEGAL_VM_DISPLAY_TYPE_IS_NOT_SUPPORTED_BY_OS >>>>>> 2017-06-19 03:28:05,030Z INFO >>>>>> [org.ovirt.engine.core.bll.exportimport.ImportVmCommand] >>>>>> (org.ovirt.thread.pool-6-thread-23) [] Lock freed to object >>>>>> 'EngineLock:{exclusiveLocks='[a >>>>>> 79e6b0e-fff4-4cba-a02c-4c00be151300=<VM, >>>>>> ACTION_TYPE_FAILED_VM_IS_BEING_IMPORTED$VmName HostedEngine>, >>>>>> HostedEngine=<VM_NAME, ACTION_TYPE_FAILED_NAME_ALREADY_USED>]', >>>>>> sharedLocks= >>>>>> '[a79e6b0e-fff4-4cba-a02c-4c00be151300=<REMOTE_VM, >>>>>> ACTION_TYPE_FAILED_VM_IS_BEING_IMPORTED$VmName HostedEngine>]'}' >>>>>> 2017-06-19 03:28:05,030Z ERROR >>>>>> [org.ovirt.engine.core.bll.HostedEngineImporter] >>>>>> (org.ovirt.thread.pool-6-thread-23) [] Failed importing the Hosted >>>>>> Engine VM >>>>>> >>>>>> The sanlock.log reports conflicts on that same host, and a >>>>>> different >>>>>> error on the other hosts, not sure if they are related. >>>>>> >>>>>> And this in the /var/log/ovirt-hosted-engine-ha/agent log on the >>>>>> host >>>>>> which I deployed the hosted engine VM on: >>>>>> >>>>>> MainThread::ERROR::2017-06-19 >>>>>> >>>>>> >>>>>> 13:09:49,743::ovf_store::124::ovirt_hosted_engine_ha.lib.ovf.ovf_store.OVFStore::(getEngineVMOVF) >>>>>> Unable to extract HEVM OVF >>>>>> MainThread::ERROR::2017-06-19 >>>>>> >>>>>> >>>>>> 13:09:49,743::config::445::ovirt_hosted_engine_ha.agent.hosted_engine.HostedEngine.config::(_get_vm_conf_content_from_ovf_store) >>>>>> Failed extracting VM OVF from the OVF_STORE volume, falling back >>>>>> to >>>>>> initial vm.conf >>>>>> >>>>>> I've seen some of these issues reported in bugzilla, but they were >>>>>> for >>>>>> older versions of oVirt (and appear to be resolved). >>>>>> >>>>>> I will install that package on the other two hosts, for which I >>>>>> will >>>>>> put them in maintenance as vdsm is installed as an upgrade. I >>>>>> guess >>>>>> restarting vdsm is a good idea after that? >>>>>> >>>>>> Thanks, >>>>>> >>>>>> Campbell >>>>>> >>>>>> On Wed, Jun 21, 2017 at 10:51 AM, Martin Sivak <msivak@redhat.com> >>>>>> wrote: >>>>>>> Hi, >>>>>>> >>>>>>> you do not have to install it on all hosts. But you should have >>>>>>> more >>>>>>> than one and ideally all hosted engine enabled nodes should >>>>>>> belong to >>>>>>> the same engine cluster. >>>>>>> >>>>>>> Best regards >>>>>>> >>>>>>> Martin Sivak >>>>>>> >>>>>>> On Wed, Jun 21, 2017 at 11:29 AM, cmc <iucounu@gmail.com> wrote: >>>>>>>> Hi Jenny, >>>>>>>> >>>>>>>> Does ovirt-hosted-engine-ha need to be installed across all >>>>>>>> hosts? >>>>>>>> Could that be the reason it is failing to see it properly? >>>>>>>> >>>>>>>> Thanks, >>>>>>>> >>>>>>>> Cam >>>>>>>> >>>>>>>> On Mon, Jun 19, 2017 at 1:27 PM, cmc <iucounu@gmail.com> wrote: >>>>>>>>> Hi Jenny, >>>>>>>>> >>>>>>>>> Logs are attached. I can see errors in there, but am unsure how >>>>>>>>> they >>>>>>>>> arose. >>>>>>>>> >>>>>>>>> Thanks, >>>>>>>>> >>>>>>>>> Campbell >>>>>>>>> >>>>>>>>> On Mon, Jun 19, 2017 at 12:29 PM, Evgenia Tokar >>>>>>>>> <etokar@redhat.com> >>>>>>>>> wrote: >>>>>>>>>> From the output it looks like the agent is down, try starting >>>>>>>>>> it by >>>>>>>>>> running: >>>>>>>>>> systemctl start ovirt-ha-agent. >>>>>>>>>> >>>>>>>>>> The engine is supposed to see the hosted engine storage domain >>>>>>>>>> and >>>>>>>>>> import it >>>>>>>>>> to the system, then it should import the hosted engine vm. >>>>>>>>>> >>>>>>>>>> Can you attach the agent log from the host >>>>>>>>>> (/var/log/ovirt-hosted-engine-ha/agent.log) >>>>>>>>>> and the engine log from the engine vm >>>>>>>>>> (/var/log/ovirt-engine/engine.log)? >>>>>>>>>> >>>>>>>>>> Thanks, >>>>>>>>>> Jenny >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> On Mon, Jun 19, 2017 at 12:41 PM, cmc <iucounu@gmail.com> >>>>>>>>>> wrote: >>>>>>>>>>> >>>>>>>>>>> Hi Jenny, >>>>>>>>>>> >>>>>>>>>>>> What version are you running? >>>>>>>>>>> >>>>>>>>>>> 4.1.2.2-1.el7.centos >>>>>>>>>>> >>>>>>>>>>>> For the hosted engine vm to be imported and displayed in the >>>>>>>>>>>> engine, you >>>>>>>>>>>> must first create a master storage domain. >>>>>>>>>>> >>>>>>>>>>> To provide a bit more detail: this was a migration of a >>>>>>>>>>> bare-metal >>>>>>>>>>> engine in an existing cluster to a hosted engine VM for that >>>>>>>>>>> cluster. >>>>>>>>>>> As part of this migration, I built an entirely new host and >>>>>>>>>>> ran >>>>>>>>>>> 'hosted-engine --deploy' (followed these instructions: >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> http://www.ovirt.org/documentation/self-hosted/chap-Migrating_from_Bare_Meta...). >>>>>>>>>>> I restored the backup from the engine and it completed >>>>>>>>>>> without any >>>>>>>>>>> errors. I didn't see any instructions regarding a master >>>>>>>>>>> storage >>>>>>>>>>> domain in the page above. The cluster has two existing master >>>>>>>>>>> storage >>>>>>>>>>> domains, one is fibre channel, which is up, and one ISO >>>>>>>>>>> domain, >>>>>>>>>>> which >>>>>>>>>>> is currently offline. >>>>>>>>>>> >>>>>>>>>>>> What do you mean the hosted engine commands are failing? >>>>>>>>>>>> What >>>>>>>>>>>> happens >>>>>>>>>>>> when >>>>>>>>>>>> you run hosted-engine --vm-status now? >>>>>>>>>>> >>>>>>>>>>> Interestingly, whereas when I ran it before, it exited with >>>>>>>>>>> no >>>>>>>>>>> output >>>>>>>>>>> and a return code of '1', it now reports: >>>>>>>>>>> >>>>>>>>>>> --== Host 1 status ==-- >>>>>>>>>>> >>>>>>>>>>> conf_on_shared_storage : True >>>>>>>>>>> Status up-to-date : False >>>>>>>>>>> Hostname : >>>>>>>>>>> kvm-ldn-03.ldn.fscfc.co.uk >>>>>>>>>>> Host ID : 1 >>>>>>>>>>> Engine status : unknown stale-data >>>>>>>>>>> Score : 0 >>>>>>>>>>> stopped : True >>>>>>>>>>> Local maintenance : False >>>>>>>>>>> crc32 : 0217f07b >>>>>>>>>>> local_conf_timestamp : 2911 >>>>>>>>>>> Host timestamp : 2897 >>>>>>>>>>> Extra metadata (valid at timestamp): >>>>>>>>>>> metadata_parse_version=1 >>>>>>>>>>> metadata_feature_version=1 >>>>>>>>>>> timestamp=2897 (Thu Jun 15 16:22:54 2017) >>>>>>>>>>> host-id=1 >>>>>>>>>>> score=0 >>>>>>>>>>> vm_conf_refresh_time=2911 (Thu Jun 15 16:23:08 2017) >>>>>>>>>>> conf_on_shared_storage=True >>>>>>>>>>> maintenance=False >>>>>>>>>>> state=AgentStopped >>>>>>>>>>> stopped=True >>>>>>>>>>> >>>>>>>>>>> Yet I can login to the web GUI fine. I guess it is not HA due >>>>>>>>>>> to >>>>>>>>>>> being >>>>>>>>>>> in an unknown state currently? Does the hosted-engine-ha rpm >>>>>>>>>>> need >>>>>>>>>>> to >>>>>>>>>>> be installed across all nodes in the cluster, btw? >>>>>>>>>>> >>>>>>>>>>> Thanks for the help, >>>>>>>>>>> >>>>>>>>>>> Cam >>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> Jenny Tokar >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> On Thu, Jun 15, 2017 at 6:32 PM, cmc <iucounu@gmail.com> >>>>>>>>>>>> wrote: >>>>>>>>>>>>> >>>>>>>>>>>>> Hi, >>>>>>>>>>>>> >>>>>>>>>>>>> I've migrated from a bare-metal engine to a hosted engine. >>>>>>>>>>>>> There >>>>>>>>>>>>> were >>>>>>>>>>>>> no errors during the install, however, the hosted engine >>>>>>>>>>>>> did not >>>>>>>>>>>>> get >>>>>>>>>>>>> started. I tried running: >>>>>>>>>>>>> >>>>>>>>>>>>> hosted-engine --status >>>>>>>>>>>>> >>>>>>>>>>>>> on the host I deployed it on, and it returns nothing (exit >>>>>>>>>>>>> code >>>>>>>>>>>>> is 1 >>>>>>>>>>>>> however). I could not ping it either. So I tried starting >>>>>>>>>>>>> it via >>>>>>>>>>>>> 'hosted-engine --vm-start' and it returned: >>>>>>>>>>>>> >>>>>>>>>>>>> Virtual machine does not exist >>>>>>>>>>>>> >>>>>>>>>>>>> But it then became available. I logged into it >>>>>>>>>>>>> successfully. It >>>>>>>>>>>>> is not >>>>>>>>>>>>> in the list of VMs however. >>>>>>>>>>>>> >>>>>>>>>>>>> Any ideas why the hosted-engine commands fail, and why it >>>>>>>>>>>>> is not >>>>>>>>>>>>> in >>>>>>>>>>>>> the list of virtual machines? >>>>>>>>>>>>> >>>>>>>>>>>>> Thanks for any help, >>>>>>>>>>>>> >>>>>>>>>>>>> Cam >>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>> Users mailing list >>>>>>>>>>>>> Users@ovirt.org >>>>>>>>>>>>> http://lists.ovirt.org/mailman/listinfo/users >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>> _______________________________________________ >>>>>>>> Users mailing list >>>>>>>> Users@ovirt.org >>>>>>>> http://lists.ovirt.org/mailman/listinfo/users >>>>> _______________________________________________ >>>>> Users mailing list >>>>> Users@ovirt.org >>>>> http://lists.ovirt.org/mailman/listinfo/users >>>> >>>> _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users