| *** kevinz has joined #openstack-zun | 00:54 | |
| *** yuanying has quit IRC | 01:12 | |
| *** yuanying has joined #openstack-zun | 01:13 | |
| *** zhurong has joined #openstack-zun | 01:13 | |
| *** yanyanhu has joined #openstack-zun | 01:31 | |
| *** zhangjl has joined #openstack-zun | 01:32 | |
| *** zhangjl1 has joined #openstack-zun | 01:51 | |
| *** zhangjl has quit IRC | 01:52 | |
| *** zhurong has quit IRC | 01:52 | |
| *** zhangjl1 has quit IRC | 01:59 | |
| *** zhangjl has joined #openstack-zun | 01:59 | |
| *** zhurong has joined #openstack-zun | 02:12 | |
| *** zhurong has quit IRC | 02:25 | |
| *** zhurong has joined #openstack-zun | 02:28 | |
| *** zhurong has quit IRC | 02:38 | |
| *** kevinz has quit IRC | 02:48 | |
| *** kevinz has joined #openstack-zun | 02:49 | |
| *** zhurong has joined #openstack-zun | 02:52 | |
| *** harlowja has quit IRC | 02:53 | |
| *** zhurong has quit IRC | 02:59 | |
| *** lakerzhou has joined #openstack-zun | 03:01 | |
| *** sudipto_ has joined #openstack-zun | 03:02 | |
| *** sudipto has joined #openstack-zun | 03:02 | |
| lakerzhou | hi~ Sudipto, regarding NUMA, I talked with a telco customer recently. They use CPU pinning without any sharing. They also use NUMA, because some VNFs request 30+ cores. | 03:18 |
|---|---|---|
| sudipto_ | lakerzhou, hi you are saying NUMA restriction is not good? | 03:19 |
| lakerzhou | we probably need to think about customer 's request. | 03:20 |
| sudipto_ | lakerzhou, it's easy to remove that restriction...but the data structure should account for NUMA. While you can ask for a non numa based cpuset if you like. | 03:21 |
| sudipto_ | So while coding, i will keep it in mind that NUMA isn't quite the restriction for dedicated... | 03:21 |
| sudipto_ | however we will support a case where we are able to even provide cpusets/memory from a NUMA node, if requested. | 03:22 |
| sudipto_ | does that sound ok to you? | 03:22 |
| lakerzhou | I am trying to confirm with you, if you know some NUMA request from NFV application. With the customer, we mainly talked about placement issue with NOVA. | 03:23 |
| sudipto_ | placement == not having enough NUMA nodes available for a request? | 03:24 |
| lakerzhou | But I don't have much insight about NUMA requirement from NFV application. | 03:24 |
| lakerzhou | they complain about nova scheduler resulting non-optimal placement, because 30+ core with NUMA setting often ends up low cpu utilization | 03:26 |
| sudipto_ | hmm...guess that problem needs further investigation. | 03:27 |
| lakerzhou | I can clarify the NUMA requirements with NFV vendor, but it may take long time to get feedback. | 03:27 |
| sudipto_ | Ok, what i am saying is - from zun, we will provide the flexibility | 03:28 |
| sudipto_ | if they want to do NUMA - they can, if they don't want to do that - they can still use it. | 03:28 |
| *** diga has quit IRC | 03:31 | |
| lakerzhou | I think there are two aspects of NUMA: 1. some applications require all CPUs from one NUMA node to gain high performance. 2. Some applications require NUMA awareness, so that all CPUs don't have to come from same NUMA node. The application has intelligence to efficiently use CPU resources. | 03:33 |
| lakerzhou | this is my understanding from reading. I can confirm this with NFV vender | 03:34 |
| sudipto_ | lakerzhou, #2 don't think applies in the container's case. | 03:35 |
| sudipto_ | #2 is when you would want to have NUMA awareness of the host's topology inside the guest. | 03:35 |
| sudipto_ | In case of containers, there isn't a topology inside the container... | 03:36 |
| lakerzhou | Anyone has issue with "zun create"? --image option is new. It seems not working with my devstack environment. | 03:37 |
| lakerzhou | sudipto_, good to know. My readings are based on VM not containers. | 03:37 |
| lakerzhou | sudipto_, are you saying there is no NUMA topology support for container? | 03:38 |
| sudipto_ | lakerzhou, you can say - a container should be on getting it's cpus from 1 NUMA node, but you can't create a NUMA topology like VMs' - experiments in the VM world have shown that if you can mock the NUMA topology of the host, inside the virtual machine - it results in better performance. But note that VMs are full fledged operating system...i am not sure if that case applies in case of containers. | 03:41 |
| lakerzhou | sudipto_, I need to read more about NUMA for containers. | 03:55 |
| *** pksingh has joined #openstack-zun | 03:58 | |
| lakerzhou | Hi pksingh~, I have a question about the new --image option in zun create. | 04:01 |
| pksingh | lakerzhou: Hello | 04:01 |
| pksingh | lakerzhou: Good Morning, yes please tell me | 04:01 |
| lakerzhou | pksingh, in my devstack environment, the newly created container stays in unknown state. | 04:02 |
| openstackgerrit | Sudipta Biswas proposed openstack/zun master: Introduce os-capability parsing https://review.openstack.org/430920 | 04:02 |
| pksingh | lakerzhou: any specific error | 04:03 |
| lakerzhou | this is command I used: zun create --name test --command "ping -c 4 8.8.8.8" -i cirros | 04:05 |
| lakerzhou | I don't see any error during creation. but let me dig more. | 04:05 |
| pksingh | lakerzhou: ok let me check if i can reproduce it | 04:05 |
| lakerzhou | pksingh: thanks, | 04:06 |
| *** hongbin_ has joined #openstack-zun | 04:08 | |
| lakerzhou | pksingh: 2017-02-09 21:55:14.888 ^[[01;31mERROR zun.image.glance.driver [^[[01;36mreq-c7c58cfc-42f0-4d30-8343-a0c96ba20a1d ^[[00;36madmin admin^[[01;31m] ^[[01;35m^[[01;31mImage cirros was not found in glance^[[00m | 04:11 |
| lakerzhou | pksingh: found the error in log, but the container state is still unknown. | 04:12 |
| pksingh | lakerzhou: give me 10 minutes, | 04:13 |
| lakerzhou | pksingh: sure. no rush | 04:13 |
| *** sudipto has quit IRC | 04:16 | |
| *** sudipto_ has quit IRC | 04:16 | |
| pksingh | lakerzhou: its working on my side | 04:19 |
| pksingh | lakerzhou: could you please paste your compute service logs | 04:19 |
| pksingh | lakerzhou: please paste at http://paste.openstack.org/ and provide the link | 04:20 |
| lakerzhou | sure, is there other command or configure change that I should run before using the image? | 04:20 |
| pksingh | lakerzhou: your command is perfectly fine | 04:21 |
| pksingh | lakerzhou: i think '-i' option is for interactive mode, its not for image. image is positional parameter, | 04:22 |
| lakerzhou | http://paste.openstack.org/show/598333/ | 04:24 |
| lakerzhou | -i <image>, --image <image> | 04:24 |
| lakerzhou | let me try again with --image | 04:24 |
| pksingh | lakerzhou: you dont need to put --image , just put image name at last | 04:27 |
| mkrai | lakerzhou: --image option is removed from zunclient | 04:27 |
| *** shu-mutou has joined #openstack-zun | 04:27 | |
| lakerzhou | mkrai: when did it removed? I got error without "-i" or "-image". | 04:31 |
| mkrai | What is the error lakerzhou ? | 04:31 |
| pksingh | lakerzhou: i think your devstack environemnt is not updated, it looks like an older one | 04:31 |
| lakerzhou | error: argument -i/--image is required | 04:32 |
| lakerzhou | let me update my devstack environment first | 04:33 |
| lakerzhou | pksingh: can you please share your local.conf? | 04:34 |
| pksingh | lakerzhou: sure | 04:34 |
| mkrai | lakerzhou: you can just update the zun only | 04:35 |
| mkrai | and restart the zun services rather then doing restack | 04:35 |
| lakerzhou | this is only config related to zun in my local.conf: enable_plugin zun https://git.openstack.org/openstack/zun | 04:37 |
| mkrai | that's enough I guess | 04:38 |
| lakerzhou | mkrai: how do I update zun, from devstack? | 04:39 |
| pksingh | lakerzhou: my conf is at http://paste.openstack.org/show/598334/ | 04:39 |
| mkrai | cd /opt/stack/zun; git pull | 04:39 |
| mkrai | and then restart zun-api and zun-compute | 04:40 |
| pksingh | lakerzhou: mkrai , i did it today. go to /opt/stack/zun, do 'git pull' and 'python setup.py install' | 04:40 |
| pksingh | mkrai: 'python setup.py install' is necessary because of image_driver bp | 04:40 |
| mkrai | yes but not if we are running it second time or so | 04:41 |
| *** adisky_ has joined #openstack-zun | 04:42 | |
| *** sudipto has joined #openstack-zun | 04:42 | |
| *** sudipto_ has joined #openstack-zun | 04:42 | |
| pksingh | i think you need to install due to https://github.com/openstack/zun/blob/master/setup.cfg#L67, if your previous code dont have this entry | 04:43 |
| lakerzhou | Thanks pksingh and mkrai | 04:43 |
| sudipto_ | pksingh, hongbin if you guys think this code: https://review.openstack.org/#/c/430920/ is alright, i will move ahead with the other patches. | 04:44 |
| pksingh | lakerzhou: please update pythonzunclient to, i think there you dont need to 'python setup.py install' | 04:44 |
| pksingh | sudipto_: will look in to it in next one hour | 04:45 |
| hongbin_ | sudipto_: LGTM | 04:45 |
| sudipto_ | hongbin_, ok pksingh sure. | 04:45 |
| *** hongbin_ has quit IRC | 04:49 | |
| *** yasemin has quit IRC | 04:57 | |
| *** diga has joined #openstack-zun | 05:01 | |
| *** janki has joined #openstack-zun | 05:07 | |
| *** jchhatbar has joined #openstack-zun | 05:29 | |
| *** janki has quit IRC | 05:30 | |
| *** yasemin has joined #openstack-zun | 05:36 | |
| *** trinaths has joined #openstack-zun | 05:58 | |
| *** jchhatbar_ has joined #openstack-zun | 06:03 | |
| *** jchhatbar has quit IRC | 06:05 | |
| *** jchhatbar_ is now known as janki | 06:18 | |
| *** pksingh has quit IRC | 07:51 | |
| *** pcastets has joined #openstack-zun | 08:09 | |
| openstackgerrit | feng.shengqin proposed openstack/python-zunclient master: Support the command "zun top" https://review.openstack.org/423974 | 08:14 |
| openstackgerrit | Sudipta Biswas proposed openstack/zun master: Introduce os-capability parsing https://review.openstack.org/430920 | 08:33 |
| *** prameswar has joined #openstack-zun | 08:38 | |
| openstackgerrit | feng.shengqin proposed openstack/python-zunclient master: Support the command "zun top" https://review.openstack.org/423974 | 08:59 |
| *** yanyanhu has quit IRC | 09:08 | |
| openstackgerrit | feng.shengqin proposed openstack/zun master: Support the command "zun top" https://review.openstack.org/432198 | 09:21 |
| openstackgerrit | Kevin Zhao proposed openstack/zun master: Add test cases for attach and resize https://review.openstack.org/432204 | 09:33 |
| *** lakerzhou has quit IRC | 09:47 | |
| openstackgerrit | Namrata proposed openstack/python-zunclient master: Add more container related test https://review.openstack.org/432221 | 10:06 |
| openstackgerrit | Namrata proposed openstack/python-zunclient master: Add OSC commands functional test https://review.openstack.org/432222 | 10:06 |
| *** kevinz has quit IRC | 10:10 | |
| *** openstackgerrit has quit IRC | 10:18 | |
| *** zhangjl has quit IRC | 10:21 | |
| *** jchhatbar has joined #openstack-zun | 10:59 | |
| *** janki has quit IRC | 11:01 | |
| *** trinaths has quit IRC | 11:23 | |
| *** jchhatbar_ has joined #openstack-zun | 11:26 | |
| *** jchhatbar has quit IRC | 11:28 | |
| *** diga has quit IRC | 11:31 | |
| *** jchhatbar has joined #openstack-zun | 11:47 | |
| *** jchhatbar is now known as janki | 11:48 | |
| *** jchhatbar_ has quit IRC | 11:50 | |
| *** sudipto has quit IRC | 12:09 | |
| *** sudipto_ has quit IRC | 12:09 | |
| *** janki has quit IRC | 12:19 | |
| *** adisky_ has quit IRC | 14:09 | |
| *** sudipto has joined #openstack-zun | 14:10 | |
| *** sudipto_ has joined #openstack-zun | 14:10 | |
| *** sudipto_ has quit IRC | 15:08 | |
| *** sudipto has quit IRC | 15:08 | |
| *** dims_ has quit IRC | 15:28 | |
| *** dims has joined #openstack-zun | 15:33 | |
| *** dims has quit IRC | 15:55 | |
| *** dims has joined #openstack-zun | 15:55 | |
| *** dims is now known as Guest12406 | 15:56 | |
| *** prameswar has quit IRC | 16:00 | |
| *** jwcroppe has joined #openstack-zun | 16:01 | |
| *** Guest12406 has quit IRC | 16:05 | |
| *** dims_ has joined #openstack-zun | 16:24 | |
| *** sudipto_ has joined #openstack-zun | 16:29 | |
| *** sudipto has joined #openstack-zun | 16:29 | |
| *** harlowja has joined #openstack-zun | 17:05 | |
| *** harlowja has quit IRC | 17:05 | |
| *** harlowja has joined #openstack-zun | 17:05 | |
| *** openstackgerrit has joined #openstack-zun | 17:12 | |
| openstackgerrit | Merged openstack/zun master: Introduce os-capability parsing https://review.openstack.org/430920 | 17:12 |
| sudipto_ | hongbin, mkrai thanks for the reviews. I will get some code written over the weekend for the generic host_capabilities methods. Also maybe will start drafting the resource provider objects. | 17:37 |
| sudipto_ | hongbin, if you could do the schema design, that would be awesome...whenever you find sometime. | 17:38 |
| *** pcastets has quit IRC | 17:47 | |
| *** sudipto has quit IRC | 18:16 | |
| *** sudipto_ has quit IRC | 18:16 | |
| openstackgerrit | Merged openstack/zun master: Updated from global requirements https://review.openstack.org/431913 | 18:35 |
| openstackgerrit | Hongbin Lu proposed openstack/zun master: Spec for Kuryr integration https://review.openstack.org/425883 | 22:21 |
| *** catintheroof has quit IRC | 23:03 | |
| *** catintheroof has joined #openstack-zun | 23:04 | |
| *** catintheroof has quit IRC | 23:08 | |
| *** jwcroppe has quit IRC | 23:34 | |
| *** jwcroppe has joined #openstack-zun | 23:35 | |
| *** jwcroppe has quit IRC | 23:39 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!