Re: [openstack-dev] [Ironic]Nova resource tracker is not able to identify resources of BM instance when fake_pxe driver used
> On Mar 29, 2016, at 08:39, Senthilprabu Shanmugavel >wrote: > > Thanks Jim. WA did the trick. Very well explained. > > Should I raise a bug for fake driver's power state?. That would be great, thanks in advance! :) // jim > >> On Tue, Mar 29, 2016 at 5:30 PM, Jim Rollenhagen >> wrote: >> On Tue, Mar 29, 2016 at 04:59:09PM +0300, Senthilprabu Shanmugavel wrote: >> > Hello, >> > >> > I am using Ironic for deploying baremetal to my openstack environment. >> > Using Liberty version on Ubuntu 14.04. I followed Ironic documentation to >> > deploy x86 servers using pxe_ipmitool. Now I have a working Ironic setup >> > for PXE boot. I also want to add my test board running ARM 64 bit CPU to >> > ironic deployment. I would like to try using fake_pxe drivers because my >> > board don't support IPMI or anything else for out of band communication. So >> > idea was to do deployment without power management, eventually fake_pxe is >> > the obvious choice. But I have problem in updating the Ironic node which I >> > will explain below. >> > >> > I created ironic node using fake_pxe driver. Added all necessary parameters >> > using node-update command. Node-show command output is given below for >> > reference >> > >> > >> > >> > Because of this during nova boot, scheduler failed to boot the BM instance. >> > >> > Can anyone help me with what's wrong in my configuration? >> >> I guess probably someone has never used the fake power driver with >> devstack. :) >> >> If a node's power state is None, it will be ignored by Nova: >> https://github.com/openstack/nova/blob/master/nova/virt/ironic/driver.py#L171 >> >> And the fake power driver doesn't set the power state to on/off itself: >> https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/fake.py#L43 >> >> We should probably fix this by changing it to: >> return task.node.power_state or states.POWER_ON >> >> In the meantime, an easy workaround would be: >> ironic node-set-power-state on >> ironic node-set-power-state off >> >> Which would have the driver 'remember' the power state is currently off, >> allowing Nova to pick up the resources. >> >> Hope that helps :) >> >> // jim >> >> > >> > >> > >> > Thanks in advance, >> > Senthil >> >> > __ >> > OpenStack Development Mailing List (not for usage questions) >> > Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe >> > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev >> >> >> __ >> OpenStack Development Mailing List (not for usage questions) >> Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe >> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev > > > > -- > Best Regards, > Senthilprabu Shanmugavel > __ > OpenStack Development Mailing List (not for usage questions) > Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ironic]Nova resource tracker is not able to identify resources of BM instance when fake_pxe driver used
Thanks Jim. WA did the trick. Very well explained. Should I raise a bug for fake driver's power state?. On Tue, Mar 29, 2016 at 5:30 PM, Jim Rollenhagenwrote: > On Tue, Mar 29, 2016 at 04:59:09PM +0300, Senthilprabu Shanmugavel wrote: > > Hello, > > > > I am using Ironic for deploying baremetal to my openstack environment. > > Using Liberty version on Ubuntu 14.04. I followed Ironic documentation > to > > deploy x86 servers using pxe_ipmitool. Now I have a working Ironic setup > > for PXE boot. I also want to add my test board running ARM 64 bit CPU to > > ironic deployment. I would like to try using fake_pxe drivers because my > > board don't support IPMI or anything else for out of band communication. > So > > idea was to do deployment without power management, eventually fake_pxe > is > > the obvious choice. But I have problem in updating the Ironic node which > I > > will explain below. > > > > I created ironic node using fake_pxe driver. Added all necessary > parameters > > using node-update command. Node-show command output is given below for > > reference > > > > > > > > Because of this during nova boot, scheduler failed to boot the BM > instance. > > > > Can anyone help me with what's wrong in my configuration? > > I guess probably someone has never used the fake power driver with > devstack. :) > > If a node's power state is None, it will be ignored by Nova: > > https://github.com/openstack/nova/blob/master/nova/virt/ironic/driver.py#L171 > > And the fake power driver doesn't set the power state to on/off itself: > > https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/fake.py#L43 > > We should probably fix this by changing it to: > return task.node.power_state or states.POWER_ON > > In the meantime, an easy workaround would be: > ironic node-set-power-state on > ironic node-set-power-state off > > Which would have the driver 'remember' the power state is currently off, > allowing Nova to pick up the resources. > > Hope that helps :) > > // jim > > > > > > > > > Thanks in advance, > > Senthil > > > > __ > > OpenStack Development Mailing List (not for usage questions) > > Unsubscribe: > openstack-dev-requ...@lists.openstack.org?subject:unsubscribe > > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev > > > __ > OpenStack Development Mailing List (not for usage questions) > Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev > -- Best Regards, Senthilprabu Shanmugavel __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ironic]Nova resource tracker is not able to identify resources of BM instance when fake_pxe driver used
On Tue, Mar 29, 2016 at 04:59:09PM +0300, Senthilprabu Shanmugavel wrote: > Hello, > > I am using Ironic for deploying baremetal to my openstack environment. > Using Liberty version on Ubuntu 14.04. I followed Ironic documentation to > deploy x86 servers using pxe_ipmitool. Now I have a working Ironic setup > for PXE boot. I also want to add my test board running ARM 64 bit CPU to > ironic deployment. I would like to try using fake_pxe drivers because my > board don't support IPMI or anything else for out of band communication. So > idea was to do deployment without power management, eventually fake_pxe is > the obvious choice. But I have problem in updating the Ironic node which I > will explain below. > > I created ironic node using fake_pxe driver. Added all necessary parameters > using node-update command. Node-show command output is given below for > reference > > > > Because of this during nova boot, scheduler failed to boot the BM instance. > > Can anyone help me with what's wrong in my configuration? I guess probably someone has never used the fake power driver with devstack. :) If a node's power state is None, it will be ignored by Nova: https://github.com/openstack/nova/blob/master/nova/virt/ironic/driver.py#L171 And the fake power driver doesn't set the power state to on/off itself: https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/fake.py#L43 We should probably fix this by changing it to: return task.node.power_state or states.POWER_ON In the meantime, an easy workaround would be: ironic node-set-power-state on ironic node-set-power-state off Which would have the driver 'remember' the power state is currently off, allowing Nova to pick up the resources. Hope that helps :) // jim > > > > Thanks in advance, > Senthil > __ > OpenStack Development Mailing List (not for usage questions) > Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev __ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
[openstack-dev] [Ironic]Nova resource tracker is not able to identify resources of BM instance when fake_pxe driver used
Hello, I am using Ironic for deploying baremetal to my openstack environment. Using Liberty version on Ubuntu 14.04. I followed Ironic documentation to deploy x86 servers using pxe_ipmitool. Now I have a working Ironic setup for PXE boot. I also want to add my test board running ARM 64 bit CPU to ironic deployment. I would like to try using fake_pxe drivers because my board don't support IPMI or anything else for out of band communication. So idea was to do deployment without power management, eventually fake_pxe is the obvious choice. But I have problem in updating the Ironic node which I will explain below. I created ironic node using fake_pxe driver. Added all necessary parameters using node-update command. Node-show command output is given below for reference ironic node-show orange-box ++-+ | Property | Value | ++-+ | chassis_uuid | | | clean_step | {} | | console_enabled| False | | created_at | 2016-03-29T12:42:53.500572+00:00 | | driver | fake_pxe | | driver_info| {u'deploy_ramdisk': u'd72d31de-8f9c-497c-8c51-87a0610f905b',| || u'deploy_kernel': u'01f16454-ef4c-44d9-ae0e-03d4286aed0a'} | | driver_internal_info | {} | | extra | {} | | inspection_finished_at | None | | inspection_started_at | None | | instance_info | {u'root_gb': 40, u'deploy_ramdisk': u'cdd02e76-13aa-4f4e- | || ba32-362119c98fa6', u'deploy_kernel': u'e1986cab- | || 9e55-4b9a-a094-2eda32361521', u'image_source': u'cdd02e76-13aa-4f4e-| || ba32-362119c98fa6'} | | instance_uuid | None | | last_error | None | | maintenance| False | | maintenance_reason | None | | name | orange-box | | power_state| None | | properties | {u'memory_mb': 2048, u'cpu_arch': u'aarch64', u'local_gb': 40, u'cpus': | || 2} | | provision_state| available | | provision_updated_at | None | | reservation| None | | target_power_state | None | | target_provision_state | None | | updated_at | 2016-03-29T13:45:07.439046+00:00 | | uuid | 8146c660-b00f-4cec-9357-a4eab714e5fd | ++-+ >From nova-compute log (with debugging enabled) 2016-03-29 15:44:02.134 19124 DEBUG nova.virt.ironic.driver [req-fbc8fc44-a3a4-43b0-bd82-cf077007320e - - - - -] Returning 1 available node(s) get_available_nodes /usr/lib/python2.7/dist-packages/nova/virt/ironic/driver.py:596 2016-03-29 15:44:02.135 19124 INFO nova.compute.resource_tracker [req-fbc8fc44-a3a4-43b0-bd82-cf077007320e - - - - -] Auditing locally available compute resources for node 8146c660-b00f-4cec-9357-a4eab714e5fd 2016-03-29 15:44:02.135 19124 DEBUG nova.virt.ironic.driver [req-fbc8fc44-a3a4-43b0-bd82-cf077007320e - - - - -] Using cache for node 8146c660-b00f-4cec-9357-a4eab714e5fd, age: 0.00126004219055 get_available_resource /usr/lib/python2.7/dist-packages/nova/virt/ironic/driver.py:621 2016-03-29 15:44:02.136 19124 DEBUG nova.compute.resource_tracker [req-fbc8fc44-a3a4-43b0-bd82-cf077007320e - - - - -] *Hypervisor: VCPU information unavailable* _report_hypervisor_resource_view /usr/lib/python2.7/dist-packages/nova/compute/resource_tracker.py:591 2016-03-29 15:44:02.137 19124 DEBUG nova.compute.resource_tracker [req-fbc8fc44-a3a4-43b0-bd82-cf077007320e - - - - -] Hypervisor/Node resource view: name=8146c660-b00f-4cec-9357-a4eab714e5fd *free_ram=0MB free_disk=0GB free_vcpus=unknown* pci_devices=None _report_hypervisor_resource_view /usr/lib/python2.7/dist-packages/nova/compute/resource_tracker.py:610 Nova resource tracker is not able to find the available resource like CPU, MEMORY and DISK and all are pointing to ZERO even though I ran "ironic node-update orange-box add properties/memory_mb=2048 properties/local_gb=40 properties/cpu_arch=aarch64 properties/cpus=2". ironic node-validate orange-box +++---+ |