| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-03-28 | |||
| 07:35:03 | priya__ | Hi team, my code breaks with "flavor not found" issue. It is seen that instance_types table is empty. The same code works in Mitaka. Is there any issue/upgrade for pike? | |
| 07:42:19 | openstackgerrit | Zhenyu Zheng proposed openstack/nova master: Add request_id to instance action notifications https://review.openstack.org/553288 | |
| 07:43:01 | priya__ | Hi team, my code breaks with "flavor not found" issue. It is seen that instance_types table is empty. The same code works in Mitaka. Is there any issue/upgrade for pike? | |
| 07:48:01 | openstackgerrit | Zhenyu Zheng proposed openstack/nova master: Add request_id to instance action notifications https://review.openstack.org/553288 | |
| 07:53:47 | Kevin_Zheng | gibi, Hi, thanks for the advice in https://review.openstack.org/553288 , but it still cannot work, so I still mocked the req_id generation call | |
| 07:54:08 | Kevin_Zheng | I put the reason in comment reply | |
| 07:56:21 | kaisers1 | mikal: ping (reg. https://review.openstack.org/#/c/554195/7/nova/privsep/libvirt.py comments from Stephen) | |
| 07:59:46 | bauzas | good morning folks | |
| 08:02:07 | bauzas | jianghuaw_: here ? | |
| 08:02:33 | bauzas | jianghuaw_: IIUC, a pGPU group is having multiple types, right? | |
| 08:03:28 | jianghuaw_ | bauzas. good morning. | |
| 08:03:29 | jianghuaw_ | yes. | |
| 08:04:23 | jianghuaw_ | XenServer supply vGPUs from GPU groups. | |
| 08:04:53 | bauzas | jianghuaw_: okay, so I don't understand your comment in https://review.openstack.org/#/c/557065/1/specs/rocky/approved/vgpu-rocky.rst@47 | |
| 08:05:15 | bauzas | how can you know which type to use for that group ? | |
| 08:05:47 | jianghuaw_ | just treat a group in xenserver as a pGPU in kvm. | |
| 08:07:08 | jianghuaw_ | the enabled_vgpu_types listed the enabled vGPU types. | |
| 08:07:35 | jianghuaw_ | we only put one vGPU type from the supported list by GPU group. | |
| 08:07:46 | jianghuaw_ | in the configure option. | |
| 08:08:24 | jianghuaw_ | If there are multiple types of PGPUs, there will be multiple GPU group. | |
| 08:09:47 | jianghuaw_ | e.g. PGPU type 1 => grp1 (support type1-1, type1-2); pGPU type 2 => grp 2 (support type2-1, type2-2); | |
| 08:10:14 | jianghuaw_ | then we set enabled_vgpu_types = type1-1, type2-1 | |
| 08:10:39 | jianghuaw_ | It's able to ensure each group will expose one type of VGPUs. | |
| 08:10:46 | bauzas | sec | |
| 08:12:53 | sahid | bauzas: i think we have an agreement here, if you can take a moment for https://review.openstack.org/#/c/511188/ | |
| 08:13:07 | bauzas | jianghuaw_: hah, that's because pGPU groups in Xen are mutually exclusive | |
| 08:13:08 | bauzas | ? | |
| 08:13:23 | bauzas | ie. you can't have both two groups sharing the same type ? | |
| 08:14:16 | bauzas | sahid: sure, it's in my radar, like I said ;) | |
| 08:14:26 | bauzas | sahid: I have to follow up on some specs :) | |
| 08:14:58 | sahid | bauzas: cool thanks | |
| 08:16:14 | jianghuaw_ | bauzas, yes. | |
| 08:16:40 | bauzas | jianghuaw_: okay, so what if the operator says enabled_vgpu_types = type1-1, type1-2 ? | |
| 08:17:28 | bauzas | jianghuaw_: taking the example of http://docs.nvidia.com/grid/5.0/grid-vgpu-user-guide/index.html#view-gpu-group-object-details-citrix-xenserver-xe | |
| 08:17:41 | jianghuaw_ | Then follow the behavior as Queens, choose the first type supported by the group. | |
| 08:18:35 | bauzas | jianghuaw_: but then it won't use the second type for a second pGPU in the same group ? | |
| 08:18:56 | jianghuaw_ | yes. | |
| 08:20:05 | jianghuaw_ | That's expected. the type in group must be predictable. | |
| 08:20:33 | jianghuaw_ | Otherwise it's not able to calculate the inventories' total. | |
| 08:21:04 | bauzas | jianghuaw_: but then it's unfortunate for the operator | |
| 08:21:19 | bauzas | because they can do something like http://docs.nvidia.com/grid/5.0/grid-vgpu-user-guide/index.html#homogeneous-grid-vgpus | |
| 08:21:29 | openstackgerrit | Yikun Jiang (Kero) proposed openstack/nova master: [WIP] Add host/hostId to instance action events API https://review.openstack.org/557145 | |
| 08:21:36 | bauzas | jianghuaw_: (first valid config) | |
| 08:29:47 | openstackgerrit | Johannes Kulik proposed openstack/nova master: Fix spelling mistake of HTTPNotFound exception https://review.openstack.org/557187 | |
| 08:32:29 | jianghuaw_ | bauzas, Do you mean we can't create different type vGPUs on different pgpu belong to a group? | |
| 08:33:07 | bauzas | jianghuaw_: a group is multiple pGPUs, right? | |
| 08:33:12 | jianghuaw_ | yes. | |
| 08:33:18 | jianghuaw_ | with same type | |
| 08:33:20 | bauzas | jianghuaw_: so, you can have one type per pGPU | |
| 08:33:38 | bauzas | jianghuaw_: no, not the same type, you just said there are multiple types per group | |
| 08:34:55 | jianghuaw_ | I mean a group can be multiple PGPUs which belong to the same PGPU type e.g. all are "Tesla M60" | |
| 08:37:06 | jianghuaw_ | Each PGPU can support multiple vGPU types; as we make group as the unit to expose vGPUs, we have to restrict each group having only one vgpu type enabled. | |
| 08:41:01 | jianghuaw_ | bauzas, with the new configure option, we can't enable multiple types in each group either. | |
| 08:43:46 | openstackgerrit | Kashyap Chamarthy proposed openstack/nova master: libvirt: Allow to specify granular CPU feature flags https://review.openstack.org/534384 | |
| 09:02:03 | kashyap | sean-k-mooney[m]: ^ Can you re-ACK it, please | |
| 09:02:46 | kashyap | (Since I moved the 'choices' stuff into config, I just removed the uppercase/lowercase check in driver.py; small mechanical change.) | |
| 09:16:34 | kashyap | imacdonn: stvnoyes1 Follow-up from last night, please update the versions of libvirt, QEMU, libguestfs for Oracle Linux here: https://wiki.openstack.org/wiki/LibvirtDistroSupportMatrix | |
| 09:17:34 | openstackgerrit | garyk proposed openstack/nova master: VMware: add log message for VIF info details https://review.openstack.org/557256 | |
| 09:18:49 | openstackgerrit | jichenjc proposed openstack/nova master: z/VM Driver: Spawn and destroy function of z/VM driver https://review.openstack.org/527658 | |
| 09:18:49 | openstackgerrit | jichenjc proposed openstack/nova master: z/VM Driver: Initial change set of z/VM driver https://review.openstack.org/523387 | |
| 09:18:50 | openstackgerrit | jichenjc proposed openstack/nova master: z/VM Driver: add power actions https://review.openstack.org/543340 | |
| 09:18:50 | openstackgerrit | jichenjc proposed openstack/nova master: z/VM Driver: add snapshot function https://review.openstack.org/534240 | |
| 09:18:51 | openstackgerrit | jichenjc proposed openstack/nova master: z/VM Driver: add get console output https://review.openstack.org/543344 | |
| 09:25:21 | bauzas | jianghuaw_: anyway, I don't want to rathole on Xen specifics | |
| 09:25:40 | bauzas | jianghuaw_: if you feel you don't need that spec, all fair, I'll make that conf opt libvirt-specific | |
| 09:28:01 | openstackgerrit | Bhagyashri Shewale proposed openstack/nova-specs master: Disallow rotation parameter 0 for 'createBackup' API https://review.openstack.org/511825 | |
| 09:28:32 | openstackgerrit | jichenjc proposed openstack/nova master: WIP: remove Ec2 object https://review.openstack.org/557150 | |
| 09:31:02 | openstackgerrit | Jie Li proposed openstack/nova-specs master: Support volume-backed server rescue https://review.openstack.org/532410 | |
| 09:32:54 | jianghuaw_ | bauzas, thanks. | |
| 09:54:17 | openstackgerrit | Bhagyashri Shewale proposed openstack/nova-specs master: Disallow rotation parameter 0 for 'createBackup' API https://review.openstack.org/511825 | |
| 09:56:34 | jianghuaw_ | bauzas, In the queens vGPU spec, we have display heads as consumable resource. Now we will make it as a trait. | |
| 09:56:35 | openstackgerrit | jichenjc proposed openstack/nova master: Move update_task_state out of try/except https://review.openstack.org/557152 | |
| 09:57:00 | jianghuaw_ | Should we just update the queens spec or should make a new spec in Rocky? | |
| 10:43:00 | openstackgerrit | Claudiu Belu proposed openstack/nova master: tests: autospecs all the mock.patch usages https://review.openstack.org/470775 | |
| 11:03:28 | openstackgerrit | Claudiu Belu proposed openstack/nova master: WIP: replace spec with autospec https://review.openstack.org/557299 | |
| 11:22:02 | openstackgerrit | Claudiu Belu proposed openstack/nova master: WIP: replace spec with autospec https://review.openstack.org/557299 | |
| 11:38:20 | priya__ | Hi team, my code breaks with "flavor not found" issue. It is seen that instance_types table is empty. The same code works in Mitaka. Is there any issue/upgrade for pike? | |
| 11:55:45 | Kevin_Zheng | priya__ seems related with cells v2 structure, | |
| 11:58:11 | Kevin_Zheng | pike reads api_db for flavor, mitaka reads nova db, so it will work with mitaka but cannot work with pike | |
| 11:58:19 | Kevin_Zheng | should perform data migration | |
| 12:00:06 | priya__ | Kevin_Zheng flavor = db.flavor_get(context, inst_type_id) is the call which breaks. how do i perform data migration? | |
| 12:02:26 | Kevin_Zheng | priya__ how did you install your nova? | |
| 12:02:44 | Kevin_Zheng | with source code or packages? | |
| 12:03:20 | Kevin_Zheng | or devstack | |
| 12:03:32 | openstackgerrit | Stephen Finucane proposed openstack/nova master: tox: Speed things up and document them https://review.openstack.org/534382 | |
| 12:03:32 | openstackgerrit | Stephen Finucane proposed openstack/nova master: tox: Make everything work with Python 3 https://review.openstack.org/556894 | |
| 12:03:33 | openstackgerrit | Stephen Finucane proposed openstack/nova master: trivial: Remove 'tools/releasenotes_tox.sh' https://review.openstack.org/534383 | |
| 12:03:44 | Kevin_Zheng | another thing is that the default flavors are removed after mitaka | |
| 12:03:55 | Kevin_Zheng | so you have to create your own flavor before use | |
| 12:25:23 | priya__ | Kevin_Zheng through package | |
| 12:25:49 | Kevin_Zheng | Then I guess you should create your own flavor before use | |
| 12:26:07 | Kevin_Zheng | because there will be no default flavors after mitaka | |
| 12:28:05 | priya__ | Kevin_Zheng yes i have created a flavor, and that shows up in nova_api,flavors table. But it is not updated in instance_types. | |
| 12:28:35 | priya__ | Kevin_Zheng flavor = db.flavor_get(context, inst_type_id) refers to instance_types table and my code breaks here..i dont understand why | |
| 12:30:09 | Kevin_Zheng | where is this code? | |
| 12:31:00 | priya__ | this is my custom code | |
| 12:31:07 | openstackgerrit | Kashyap Chamarthy proposed openstack/nova master: libvirt: Allow to specify granular CPU feature flags https://review.openstack.org/534384 | |
| 12:31:22 | kashyap | alex_xu_: Thanks for the eagle eyes; I think '20' is the lucky number :-) | |
| 12:31:35 | Kevin_Zheng | OK then you access the wrong db | |
| 12:31:57 | Kevin_Zheng | because in Pike the flavor is not in that table anymore | |