| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2017-01-12 | |||
| 22:16:09 | openstackgerrit | Monty Taylor proposed openstack/python-openstackclient: Install from /opt/stack/new instead of git.o.o https://review.openstack.org/419670 | |
| 22:21:47 | mordred | stevemar: left comment on the project-config patch - I think the one above ^^ is also needed | |
| 23:43:28 | adriant | I seem to be getting a devstack failure because of this: http://paste.openstack.org/show/594785/ | |
| 23:43:38 | adriant | is this a known problem? Or something I can fix? | |
| 23:44:01 | openstackgerrit | Merged openstack/python-openstackclient: unskip network qos rule type functional test https://review.openstack.org/419129 | |
| 23:44:02 | adriant | Clean VM ubuntu 14.04 vm, mitaka devstack. | |
| 23:45:41 | adriant | I can't find anything special that I'm doing differently, and I've been able to build devstack for months just fine. :( | |
| 23:46:31 | dtroyer | adriant: what are the versions of python-openstackclient, python-openstacksdk, and os-client-config? | |
| 23:46:43 | adriant | I'll do a pip freeze and check | |
| 23:47:18 | adriant | http://paste.openstack.org/show/594786/ | |
| 23:47:57 | adriant | openstacksdk==0.8.1, python-openstackclient==2.3.0 | |
| 23:48:18 | adriant | this is what devstack at "stable/mitaka" is installing | |
| 23:48:32 | adriant | os-client-config==1.16.0 | |
| 23:52:05 | dtroyer | where is that traceback coming from? is this the first osc comamnd being run? | |
| 23:52:25 | adriant | from running stack.sh | |
| 23:52:34 | adriant | it's from just before the build exited | |
| 23:52:44 | dtroyer | so the cloud is up? | |
| 23:52:53 | adriant | nope | |
| 23:53:09 | dtroyer | what is the osc command? | |
| 23:53:14 | adriant | well... sec | |
| 23:53:17 | adriant | it's the image uploading | |
| 23:53:33 | adriant | so yes the cloud probably is up in some basic capacity. | |
| 23:54:33 | dtroyer | I'm wondering about the sdk version, and why it fails now | |
| 23:54:52 | dtroyer | we only use the SDK for network commands, even today, so it has to be related to that somehow | |
| 23:55:10 | adriant | hmmm | |
| 23:55:16 | adriant | no the devstack build is nowhere near done | |
| 23:55:37 | adriant | it died in the service setup stages for glance it seems. | |
| 23:55:37 | dtroyer | but if even just keystone is running osc is functioning at some level | |
| 23:55:53 | adriant | I'll reinstall OSC with a later version | |
| 23:55:58 | adriant | and try talking to at elast keystone | |
| 23:56:22 | adriant | I can possibly install OSC myself, and then run devstack | |
| 23:56:24 | dtroyer | you'll get dependency conflicts if you don't put it into a venv | |
| 23:57:34 | dtroyer | the traceback indicates an assumption about the sdk, I have no idea when openstack.__verion__ was added (or removed) | |
| 23:58:09 | adriant | Would you build a devstack for mitaka of your own and see if you get the same issue? | |
| 23:58:44 | adriant | I'm vaguely curious if this is somewhat ubuntu related. | |
| 23:58:45 | dtroyer | maybe tomorrow. my devstack vm resources are tied up with neutron-related build failures right now | |
| 23:59:02 | adriant | and can talk to keystone with newer version of OSC | |
| 23:59:30 | dtroyer | current OSc should be able to do everything you need, except install in the same system as a running mitaka devstack | |
| 23:59:30 | adriant | checking as to what is actually in the catalog now | |
| 23:59:57 | adriant | Oh yeah, I assume as much, my problem is that devstack itself is failing to finish | |
| #openstack-sdks - 2017-01-13 | |||
| 00:00:13 | adriant | the bash uses the openstack command to upload the glance images it seems | |
| 00:00:42 | dtroyer | did Glance (or anyone actually) do a stable release recently? this feels like there is a constraint not being honored and the wrong SDK is present | |
| 00:01:51 | adriant | last glance commit to stable/mitaka was 28 sep 2016 | |
| 00:01:55 | adriant | so I doubt that's the issue here. | |
| 00:01:56 | dtroyer | it is end of day for me, but a couple of folk who I know use Mitaka clouds will be around soon (/me waves at RuiChen) and may be able to confirm the rigth versions of sdk and friends for you | |
| 00:02:11 | adriant | thanks :) | |
| 00:02:17 | adriant | I'll dig into it myself a bit more | |
| 00:02:29 | adriant | maybe trying install it in 16.04 instead | |
| 00:03:11 | dtroyer | I haven't run mitaka devstack on 16.04, not sure we ever tested that actually, I think we started with newton. good luck | |
| 00:03:41 | adriant | *shrug* | |
| 00:04:33 | adriant | I just need it working, even if that means editing random requirements myself to let stuff pass | |
| 00:05:32 | openstackgerrit | Merged openstack/python-openstackclient: Fix quota show output https://review.openstack.org/419067 | |
| 02:29:40 | openstackgerrit | Steve Martinelli proposed openstack/python-openstackclient: Install from /opt/stack/new instead of git.o.o https://review.openstack.org/419670 | |
| 03:04:09 | RuiChen | hi stevemar could you reopen the backport patch https://review.openstack.org/#/c/354271/ ? it's that SSL bug which I talk with dtroyer , it's valuable for stable branch. | |
| 03:25:27 | stevemar | RuiChen: ack | |
| 03:26:01 | stevemar | RuiChen: lets see if jenkins likes it | |
| 03:31:21 | RuiChen | stevemar : ) | |
| 03:40:52 | openstackgerrit | Steve Martinelli proposed openstack/python-openstackclient: Run functional tests with SDK 0.9.12 https://review.openstack.org/418650 | |
| 05:10:43 | stevemar | RuiChen: if you see huanxuan let him know i got the tests running on SDK 0.9.12, looks like there are a few failures still >.< | |
| 05:11:00 | stevemar | https://review.openstack.org/#/c/418650/ and the results -> http://logs.openstack.org/50/418650/9/check/gate-osc-dsvm-functional-ubuntu-xenial/dae2a65/testr_results.html.gz | |
| 06:46:40 | RuiChen | stevemar: ok, I will try to notify him by wechat, actually we are in different company and different city, but never mind, I will try :) | |
| 06:47:16 | openstackgerrit | Qiming Teng proposed openstack/python-openstacksdk: Add cluster_operation and node_operation https://review.openstack.org/419783 | |
| 06:59:00 | RuiChen | stevemar: aohuanxuan is in business trip, I can try to toke on the issue https://review.openstack.org/#/c/418650/ , not familiar it , might need some time | |
| 07:42:06 | RuiChen | hi reedip, do you work on the bug https://bugs.launchpad.net/python-openstackclient/+bug/1655445 , I think it related with quota update failed that stevemar mentioned in https://review.openstack.org/#/c/418650/ | |
| 07:42:06 | openstack | Launchpad bug 1655445 in python-openstackclient "Cannot update subnetpool quota" [Undecided,New] - Assigned to Reedip (reedip-banerjee) | |
| 07:42:33 | reedip | RuiChen : its in my ToDo today | |
| 07:43:24 | reedip | RuiChen : I have 0.9.12 ( I like living on the edge :P ) | |
| 07:43:56 | reedip | RuiChen : so would be fixing that in the Openstack SDK itseld, I talked with briancurtin about it earlier this week | |
| 07:44:47 | RuiChen | seems neutron server side don't accept id when update quota | |
| 07:44:53 | reedip | yes, it doesnt | |
| 07:44:55 | RuiChen | https://www.irccloud.com/pastebin/kYQKtf7A/ | |
| 07:45:16 | reedip | thats what I mentioned in the bug | |
| 07:46:35 | reedip | I have made a few changes, but need to modify resource2 in openstack SDK | |
| 07:48:58 | RuiChen | yes, you are right, the object id is added common logic of resource2 | |
| 07:50:10 | reedip | RuiChen : The fix is simple , ideally | |
| 07:50:23 | reedip | RuiChen: quota update uses proxy | |
| 07:51:28 | reedip | From openstack/network/v2/_proxy.py | |
| 07:52:08 | reedip | basically we need to send a check from here NOT to check ID required in openstack/proxy.py | |
| 07:53:07 | reedip | But to do that we need to overwrite the _update function in proxy | |
| 07:53:36 | reedip | or send an additional attribute from update_quota function which is popped out in the _update function | |
| 07:54:24 | RuiChen | thank you, I'm thinking about why neutron side don't accept the id attribute? looks like it break API forward compatibility | |
| 07:55:03 | reedip | RuiChen : Neutron API does not need ID as an attribute | |
| 07:55:27 | reedip | thats because it treats quotas as a resource, and the ID acts an end point to that resource | |
| 07:56:03 | reedip | if you try and update the quota using : "neutron --debug quota-update --subnet 1" , you will see the resource end point | |
| 07:56:24 | reedip | considering you have the neutronclient installed in your system | |
| 07:56:38 | RuiChen | let me try, thank your clarification :) | |
| 07:57:15 | reedip | NP :) | |
| 07:59:04 | RuiChen | https://www.irccloud.com/pastebin/JeIIAoRM/ | |
| 08:01:52 | RuiChen | reedip: so neutron don't accept id always, openstacksdk shouldn't add id, right? | |
| 08:02:24 | reedip | RuiChen : Neutron doesnt accept IDs for only certain conditions | |
| 08:02:27 | reedip | like quota | |
| 08:02:38 | reedip | maybe this would change in the future releases, but not now | |
| 08:02:54 | reedip | openstacksdk should not add ID for specific conditions | |
| 08:03:03 | RuiChen | get it, thank you :) | |
| 08:03:28 | reedip | RuiChen : what would be interesting is quota list for a specific tenant :D | |
| 08:05:30 | reedip | Like when I ran it for a specific project ( demo ) , neutron evaluated the REST request as : | |
| 08:05:33 | reedip | neutronclient.v2_0.client GET call to neutron for http://10.0.4.186:9696/v2.0/quotas/062fe3fccc174db385b1a35a1ee3152f.json | |
| 08:05:37 | RuiChen | I'm not sure if all the services support to list specified tenant quota | |
| 08:05:51 | reedip | RuiChen: neutron does , in neutronclient | |
| 08:06:02 | reedip | then it shouldnt avoid doing the same in openstackclient :D | |