| openstackgerrit | Hongbin Lu proposed openstack/zun master: [WIP] Add compute node to DB and objects https://review.openstack.org/436714 | 00:12 |
|---|---|---|
| *** hongbin has quit IRC | 00:13 | |
| *** yuanying_ has joined #openstack-zun | 00:58 | |
| *** yuanying has quit IRC | 01:02 | |
| *** kevinz has joined #openstack-zun | 01:11 | |
| *** yanyanhu has joined #openstack-zun | 01:22 | |
| *** yuanying_ has quit IRC | 02:17 | |
| openstackgerrit | Wenzhi Yu proposed openstack/zun master: [WIP]Introduce container composition https://review.openstack.org/437759 | 02:21 |
| *** Wenzhi has joined #openstack-zun | 02:21 | |
| *** yuanying has joined #openstack-zun | 02:34 | |
| *** caowei has joined #openstack-zun | 02:43 | |
| *** yuanying has quit IRC | 02:45 | |
| *** hongbin has joined #openstack-zun | 02:52 | |
| *** hongbin has quit IRC | 03:20 | |
| *** pksingh has joined #openstack-zun | 03:31 | |
| openstackgerrit | Merged openstack/python-zunclient master: Improve error message parsing https://review.openstack.org/437225 | 03:39 |
| *** kevinz has quit IRC | 04:04 | |
| *** yuanying has joined #openstack-zun | 04:06 | |
| *** diga has joined #openstack-zun | 04:34 | |
| *** janki has joined #openstack-zun | 05:24 | |
| *** sudipto has joined #openstack-zun | 05:28 | |
| *** sudipto_ has joined #openstack-zun | 05:28 | |
| *** jwcroppe_ has quit IRC | 05:45 | |
| *** yanyanhu has quit IRC | 05:46 | |
| *** yanyanhu has joined #openstack-zun | 06:06 | |
| *** pksingh has quit IRC | 06:06 | |
| *** pksingh has joined #openstack-zun | 06:31 | |
| *** diga has quit IRC | 06:42 | |
| *** jwcroppe has joined #openstack-zun | 06:47 | |
| *** jwcroppe has quit IRC | 06:51 | |
| *** pksingh has quit IRC | 07:11 | |
| *** pksingh has joined #openstack-zun | 07:41 | |
| *** pksingh has quit IRC | 07:41 | |
| *** pksingh has joined #openstack-zun | 07:41 | |
| *** pksingh has quit IRC | 07:42 | |
| *** jwcroppe has joined #openstack-zun | 07:48 | |
| *** diga has joined #openstack-zun | 07:50 | |
| *** jwcroppe has quit IRC | 07:53 | |
| openstackgerrit | feng.shengqin proposed openstack/zun master: Support the command "zun diff" https://review.openstack.org/437827 | 08:03 |
| *** Wenzhi has quit IRC | 08:56 | |
| openstackgerrit | feng.shengqin proposed openstack/python-zunclient master: Support the command "zun diff" https://review.openstack.org/437843 | 09:00 |
| *** anteaya has quit IRC | 09:27 | |
| *** anteaya has joined #openstack-zun | 09:30 | |
| *** yuanying has quit IRC | 09:42 | |
| *** caowei has quit IRC | 10:09 | |
| openstackgerrit | Madhuri Kumari proposed openstack/zun master: Remove 'bay' terminology from source code https://review.openstack.org/437866 | 10:10 |
| *** bhavani has quit IRC | 10:33 | |
| *** yuanying has joined #openstack-zun | 10:42 | |
| *** yanyanhu has quit IRC | 10:49 | |
| *** jwcroppe has joined #openstack-zun | 10:51 | |
| *** jwcroppe has quit IRC | 10:56 | |
| *** sudipto has quit IRC | 11:18 | |
| *** sudipto_ has quit IRC | 11:18 | |
| *** trinaths has joined #openstack-zun | 11:49 | |
| *** trinaths1 has joined #openstack-zun | 11:51 | |
| *** trinaths has quit IRC | 11:54 | |
| *** trinaths1 has quit IRC | 12:13 | |
| *** trinaths has joined #openstack-zun | 12:13 | |
| *** openstackgerrit has quit IRC | 12:18 | |
| *** sudipto_ has joined #openstack-zun | 12:50 | |
| *** sudipto has joined #openstack-zun | 12:50 | |
| *** diga has quit IRC | 13:24 | |
| *** kevinz has joined #openstack-zun | 13:32 | |
| *** sudipto_ has quit IRC | 13:37 | |
| *** sudipto has quit IRC | 13:37 | |
| *** kevinz has quit IRC | 13:42 | |
| *** kevinz has joined #openstack-zun | 13:43 | |
| *** trinaths has quit IRC | 13:46 | |
| *** jchhatbar has joined #openstack-zun | 13:52 | |
| *** janki has quit IRC | 13:52 | |
| *** jwcroppe has joined #openstack-zun | 13:54 | |
| *** jwcroppe has quit IRC | 13:58 | |
| *** jwcroppe has joined #openstack-zun | 14:09 | |
| *** sudipto_ has joined #openstack-zun | 14:42 | |
| *** sudipto has joined #openstack-zun | 14:42 | |
| *** kevinz has quit IRC | 14:46 | |
| *** jwcroppe has quit IRC | 14:47 | |
| *** jwcroppe has joined #openstack-zun | 14:48 | |
| *** jwcroppe has quit IRC | 14:53 | |
| *** sudipto has quit IRC | 14:56 | |
| *** sudipto_ has quit IRC | 14:56 | |
| *** sudipto has joined #openstack-zun | 14:59 | |
| *** sudipto_ has joined #openstack-zun | 14:59 | |
| *** sudipto has quit IRC | 15:01 | |
| *** sudipto_ has quit IRC | 15:01 | |
| *** jchhatbar has quit IRC | 15:06 | |
| *** hongbin has joined #openstack-zun | 15:07 | |
| *** sudipto has joined #openstack-zun | 15:12 | |
| *** sudipto_ has joined #openstack-zun | 15:12 | |
| *** sudipto has quit IRC | 15:25 | |
| *** sudipto_ has quit IRC | 15:25 | |
| *** pksingh has joined #openstack-zun | 16:04 | |
| *** jwcroppe has joined #openstack-zun | 16:11 | |
| pksingh | hongbin: good morning, r u here? | 16:23 |
| hongbin | pksingh: yes | 16:23 |
| pksingh | hongbin: when i create container using nova driver, i am not able to ping the container ip | 16:24 |
| hongbin | pksingh: if you assign a floating ip to the infra container, are you able to ping again? | 16:24 |
| pksingh | hongbin: when i assign floating ip to sandbox, i am able to ping the floating ip but not the other ip | 16:25 |
| hongbin | pksingh: yes, same here | 16:26 |
| pksingh | hongbin: even route is not listing the network | 16:26 |
| *** openstackgerrit has joined #openstack-zun | 16:27 | |
| openstackgerrit | Merged openstack/zun master: Remove 'bay' terminology from source code https://review.openstack.org/437866 | 16:27 |
| pksingh | hongbin: have you installed multinode zun | 16:28 |
| hongbin | pksingh: no, i haven't | 16:28 |
| pksingh | hongbin: ok, for the service BP, i was thinking if we assign the floating IP to sanbox and access the container through flotingip:port , is it OK? | 16:29 |
| hongbin | pksingh: it might be fine, but i feel we need another name if we implemented it in this way | 16:31 |
| pksingh | hongbin: could you please elaborate? | 16:32 |
| hongbin | pksingh: perhaps, it should not called "service", it should be zun expose or something else | 16:32 |
| pksingh | hongbin: so when there is service then we need to loadbalance it, right? | 16:32 |
| hongbin | pksingh: in the future, yes | 16:33 |
| pksingh | hongbin: what is difference between 'service' and 'expose'? | 16:33 |
| hongbin | pksingh: per my understanding, service is basically a load balancer | 16:33 |
| hongbin | pksingh: expose is to expose a port | 16:33 |
| pksingh | hongbin: yes, i was thinking the same, so when we have concept of replication controller then we need loadbalancing, right? | 16:34 |
| hongbin | pksingh: yes | 16:34 |
| hongbin | pksingh: even without replication controller, if the container is die and recreated, the ip address should be stay the same | 16:35 |
| pksingh | hongbin: at that time can we create a neutron loadbalancer on the top of all the fips assigned to different sandboxes? | 16:35 |
| hongbin | (the ip address of service) | 16:35 |
| hongbin | pksingh: i think it is better to create a neutron load balancer and assign fip to the load balancer | 16:36 |
| pksingh | hongbin: and this loadbalancer will move the traffic to internal ips of the sandbox? | 16:37 |
| hongbin | pksingh: yes | 16:37 |
| pksingh | hongbin: ok, and fip does not change right? | 16:37 |
| hongbin | pksingh: load balancer itself will have an internal ip, and i expect it won't change (unless the load balancer die) | 16:38 |
| hongbin | pksingh: and the fip associated to the load balancer should be the same | 16:38 |
| pksingh | hongbin: "should be same", same to what? | 16:39 |
| hongbin | pksingh: if a container die and re-create, the ip address of neutron load balancer should be the same as before | 16:39 |
| pksingh | hongbin: yes, i agree | 16:40 |
| pksingh | hongbin: does kuryr provide apis to create floating IP, if the sandbox is created through docker driver? | 16:40 |
| hongbin | pksingh: afaik, it doesn't | 16:41 |
| hongbin | pksingh: perhaps, they will support "docker expose" by tuning the security group of the private network, but i am not sure about that | 16:42 |
| pksingh | hongbin: i was reading some where in kuryr kubernetes and i think there is some provision, but not sure about kuryr ? | 16:43 |
| hongbin | pksingh: yes, kuryr-kubernetes support external load balancer, but not for kuryr-libnetwork | 16:43 |
| pksingh | hongbin: yes, so can we implement 'zun expose' for nova driver as of now? | 16:44 |
| hongbin | pksingh: i guess we can | 16:44 |
| pksingh | hongbin: is this approach seams genuine? | 16:44 |
| hongbin | pksingh: i haven't given it more thoughts, but it seems the right feature | 16:45 |
| hongbin | pksingh: that is my opinion only | 16:45 |
| pksingh | hongbin: yes obiously, i can propose a spec and take the view point of all the people | 16:46 |
| hongbin | pksingh: yes, that would be great | 16:46 |
| pksingh | hongbin: floating ip approach would not be conflicting with compute node's network, which we discussed earlier for exposing the container port? | 16:47 |
| hongbin | pksingh: i guess the expose is for tuning the container's security group to expose a port? if yes, then it should be fine | 16:48 |
| hongbin | pksingh: then, i think it is better to create a dedicated security group per container by default (unless users explicitely specify a security group) | 16:49 |
| pksingh | hongbin: i was thinking like this, assign the floating IP, tune the security group through nova api calls | 16:49 |
| pksingh | hongbin: yes i agree to your point of creating a dedicated security group | 16:50 |
| hongbin | pksingh: i don't think floating ip is supported in all clouds | 16:50 |
| hongbin | pksingh: i knew a lot of clouds don't have floating ip at all | 16:51 |
| hongbin | pksingh: so the creation of floating ip is better to be optional or a separated command | 16:51 |
| pksingh | hongbin: then we just need to tune the security group | 16:51 |
| hongbin | pksingh: yes | 16:51 |
| pksingh | hongbin: and access that through the interalIP:port? | 16:51 |
| hongbin | pksingh: yes, (although our devstack cannot do that correctly) | 16:52 |
| pksingh | hongbin: ok, any idea why devstack is not doing correctly? | 16:53 |
| hongbin | pksingh: i will try to figure out why we cannot ping the internal ip | 16:53 |
| pksingh | hongbin: i will look into it too | 16:53 |
| hongbin | pksingh: no idea for now, need to look into it, possibly there are some setting in the devstack to tune it | 16:53 |
| hongbin | pksingh: thx | 16:53 |
| pksingh | hongbin: ok, np | 16:53 |
| pksingh | hongbin: today is holiday for you? | 16:53 |
| hongbin | pksingh: no | 16:54 |
| hongbin | pksingh: it is on the working hour, so you find me at the right time | 16:54 |
| pksingh | hongbin: is it friday or saturday? | 16:54 |
| hongbin | friday morning | 16:54 |
| pksingh | hongbin: ok great, we are off here, some festive holiday , so long weekend :) | 16:55 |
| hongbin | pksingh: good for you :) | 16:55 |
| pksingh | hongbin: yup, ok now going to look into the work, ping you latter if required any help | 16:56 |
| hongbin | pksingh: sure | 16:57 |
| pksingh | hongbin: bbye, have a good day ahead :) | 16:57 |
| hongbin | pksingh: thx | 16:57 |
| *** sudipto_ has joined #openstack-zun | 16:57 | |
| *** sudipto has joined #openstack-zun | 16:57 | |
| pksingh | sudipto: at this time | 16:58 |
| pksingh | :) | 16:58 |
| sudipto_ | yeah :) what about you ? | 16:58 |
| pksingh | sudipto_: in past couple of weeks couldn't focus much on zun due to some work, now trying to start the coding again | 16:59 |
| pksingh | sudipto_: you are doing great :) | 16:59 |
| sudipto_ | pksingh, great :) | 16:59 |
| sudipto_ | I try to divide my time too :) | 16:59 |
| pksingh | sudipto: yes | 17:00 |
| pksingh | sudipto_: any fun plan for this weekend? | 17:00 |
| pksingh | sudipto_: was today off for you? | 17:00 |
| sudipto_ | pksingh, nothing much in plan right now. No it was not a holiday today... | 17:01 |
| sudipto_ | Was it for you? | 17:01 |
| pksingh | sudipto_: yes, it was off for us | 17:01 |
| pksingh | sudipto_: shivratri | 17:01 |
| sudipto_ | pksingh, that's great. | 17:01 |
| pksingh | yes | 17:02 |
| sudipto_ | Generally saw a lot less traffic on the roads - so i guess many offices were shut. | 17:02 |
| pksingh | sudipto_: yes i think so | 17:02 |
| pksingh | sudipto_: office in manyta tech park ? | 17:02 |
| sudipto_ | pksingh, no EGL :) | 17:03 |
| pksingh | sudipto_: ok | 17:03 |
| pksingh | sudipto_: no idea about EGL, i relocated to banglore very recently | 17:03 |
| *** jwcroppe has quit IRC | 17:14 | |
| openstackgerrit | Hongbin Lu proposed openstack/zun master: [WIP] Add compute node to DB and objects https://review.openstack.org/436714 | 17:15 |
| pksingh | hongbin: i installed devstack today, and i am not able to create second container after creating first one | 17:42 |
| pksingh | hongbin: getting error "Went to status ERROR due to "Message: No valid host was found. There are not enough hosts available., Code: 500" | 17:42 |
| pksingh | hongbin: here is the error http://paste.openstack.org/show/600407/ | 17:44 |
| pksingh | sudipto: ^^ any idea | 17:45 |
| pksingh | hongbin: sudipto_ http://paste.openstack.org/show/600411/ | 17:49 |
| sudipto | pksingh, I am not quite familiar with the sandbox code, but it sounds like the sandbox creation didn't quite go through as expected. | 17:56 |
| pksingh | sudipto_: yes, but not sure why nova is not finding the resources | 17:56 |
| sudipto | Are we trying to spawn A VM? as a part of the sandbox creation? | 17:56 |
| pksingh | sudipto_: may be some problem with my vagrant box | 17:56 |
| sudipto | I thought it was some kind of a paused container? NO? | 17:56 |
| sudipto | Sorry i haven't played with the sandbox feature at all. | 17:56 |
| pksingh | sudipto_: its conatiner | 17:57 |
| pksingh | sudipto_: paused conatiner | 17:57 |
| sudipto | Phew, your logs state things as VM :) So i was confused. | 17:57 |
| pksingh | sudipto_: we are using nova docker for creating sandbox | 17:57 |
| pksingh | sudipto_: thats why logs are similar to VM | 17:57 |
| sudipto | Ah i see, the older nomenclature. | 17:57 |
| pksingh | i will debug more | 17:58 |
| *** sudipto has quit IRC | 18:03 | |
| *** sudipto_ has quit IRC | 18:03 | |
| *** sudipto has joined #openstack-zun | 18:03 | |
| *** sudipto has quit IRC | 18:03 | |
| hongbin | pksingh: possibly, you host doesn't have enough disk/cpu/memory | 18:39 |
| hongbin | pksingh: the sandbox instance takes m1.small by default | 18:40 |
| pksingh | hongbin: yes, it seems vagrant box just got one vcpu, and thats consumed by 1st sandbox | 18:40 |
| pksingh | hongbin: i am creating vagrant box again, | 18:40 |
| pksingh | hongbin: thnx | 18:40 |
| pksingh | hongbin: why you have choosen the sandbox flavour m1.small? | 18:51 |
| pksingh | hongbin: any specific reason? | 18:51 |
| hongbin | pksingh: i just need to hardcode a flavor | 18:52 |
| hongbin | pksingh: there is no specific reason | 18:53 |
| pksingh | hongbin: i think the image kubernetes/pause is doing nothing, so tiny would be sufficient? | 18:53 |
| hongbin | pksingh: yes, feel free to change that | 18:53 |
| pksingh | hongbin: i think if the user's container needs more memory, it would not impact sandbox container anyhow? | 18:54 |
| hongbin | pksingh: i don't think it would | 18:55 |
| hongbin | pksingh: right now, sandbox and user container are on a different cgroup, so their resources are independent | 18:56 |
| pksingh | hongbin: any plan in future when their cgroups can be same? | 18:56 |
| hongbin | pksingh: i was thinking about that, it needs a bp to propose the idea and go through a team discussion | 18:57 |
| pksingh | hongbin: what is the basic idea could you please tell, just summary would be enough | 18:58 |
| hongbin | here is what i thought | 18:59 |
| hongbin | 1. the sandbox container create the parent cgroup for its user container | 18:59 |
| hongbin | 2. user container join the cgroup by using the option --cgroup-parent | 19:00 |
| pksingh | hongbin: ok, | 19:01 |
| hongbin | this is an alternative idea, i am not sure if it is a good idea or not, but k8s implemented pod in this way | 19:01 |
| pksingh | hongbin: so sandbox will have momory equal to sum of its containers? | 19:02 |
| hongbin | pksingh: yes | 19:02 |
| pksingh | hongbin: so by tracking sandbox resource usage we can track the total resource usgae of the all the containers inside it? | 19:03 |
| hongbin | pksingh: yes | 19:03 |
| pksingh | hongbin: so i think that is going to over resource utilization in nova, due to fix flavour size | 19:04 |
| hongbin | pksingh: over resource utilization? | 19:04 |
| pksingh | means if we need 3GB total and 3GB flavour is not there in nova, so we have to select flavour which have more resource than 3GB, right? | 19:05 |
| hongbin | yes, that is the drawback | 19:05 |
| pksingh | hongbin: but may be we dont need nova driver once kuryr solves all our use cases :) | 19:06 |
| hongbin | another option is leave it as is (let each user container choose the resource constraints) | 19:06 |
| hongbin | pksingh: perhaps it might not | 19:07 |
| pksingh | hongbin: hmm | 19:07 |
| pksingh | hongbin: i like first approach | 19:07 |
| hongbin | pksingh: first approach is? | 19:07 |
| pksingh | that would be easy for tracking and scheduling and all | 19:07 |
| pksingh | cgroups one | 19:07 |
| hongbin | i see | 19:08 |
| hongbin | then, i can file a bp about that | 19:08 |
| pksingh | hongbin: yes, i think we can go ahead and decide by discussion | 19:08 |
| hongbin | ok | 19:08 |
| pksingh | hongbin: btw, i am able to ping each conatiner's IP from inside the containers, but not from outside | 19:09 |
| pksingh | hongbin: isn't it neutron's feature? | 19:09 |
| pksingh | hongbin: private networks can be accessed only from inside the neutron network | 19:10 |
| hongbin | pksingh: yes, it seems it is | 19:11 |
| pksingh | hongbin: i dont know about the neutron at all, but someone was saying we can access them using flat networks but not sure | 19:12 |
| hongbin | pksingh: it seems the flat network didn't work in our devstack as well? | 19:13 |
| pksingh | hongbin: yes | 19:13 |
| hongbin | pksingh: ok, then it seems there are some issues on our devstack | 19:14 |
| pksingh | hongbin: not sure, but it seems it is | 19:15 |
| hongbin | that needs to be fixed, although not a high priority | 19:16 |
| pksingh | hongbin: yes | 19:16 |
| hongbin | possibly a low priority | 19:16 |
| pksingh | will try to find the cause | 19:16 |
| hongbin | thx | 19:18 |
| hongbin | pksingh: created a bp for that: https://blueprints.launchpad.net/zun/+spec/resource-management | 19:26 |
| *** pksingh has quit IRC | 19:26 | |
| *** jwcroppe has joined #openstack-zun | 20:23 | |
| *** jwcroppe has quit IRC | 20:44 | |
| *** lakerzhou has joined #openstack-zun | 21:00 | |
| lakerzhou | Hi There, anyone online. I am testing zun api now. {{IPADDR}}:9517/v1 returns nothing. Anyone can give me an example so I can dig more by looking at the codes. thanks | 21:01 |
| *** hongbin_ has joined #openstack-zun | 21:02 | |
| lakerzhou | BTW, I am using postman, it works fine with other APIs, for example nova | 21:02 |
| lakerzhou | Hi Hongbin, can you give me a URL example of zun API? | 21:03 |
| *** hongbin has quit IRC | 21:04 | |
| *** catinthe_ has joined #openstack-zun | 21:25 | |
| *** catintheroof has quit IRC | 21:27 | |
| *** catintheroof has joined #openstack-zun | 21:27 | |
| *** catinthe_ has quit IRC | 21:31 | |
| *** catintheroof has quit IRC | 21:33 | |
| *** lakerzhou has quit IRC | 22:03 | |
| *** jwcroppe has joined #openstack-zun | 22:05 | |
| *** jwcroppe_ has joined #openstack-zun | 22:06 | |
| *** hongbin_ has quit IRC | 22:08 | |
| *** hongbin has joined #openstack-zun | 22:09 | |
| *** jwcroppe has quit IRC | 22:10 | |
| openstackgerrit | Hongbin Lu proposed openstack/zun master: Add compute node to DB and objects https://review.openstack.org/436714 | 22:41 |
| openstackgerrit | Hongbin Lu proposed openstack/zun master: Add wsgi script file and sample config https://review.openstack.org/437190 | 23:41 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!