| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2022-08-22 | |||
| 12:13:24 | gtema | ok, will have a look later today | |
| 12:13:54 | jm1 | gtema: updating the server fails because no floating ip gets attached when auto_ip is set to true | |
| 12:14:33 | gtema | uhg, that is one of the most tricky cases: how IP is getting assinged. There are 2000000000 different ways | |
| 12:14:42 | gtema | will have a deeper look | |
| 12:17:56 | anskiy | gtema: could you, please, take a look at https://review.opendev.org/c/openstack/openstacksdk/+/848095/ again? | |
| 12:18:20 | gtema | ok | |
| 12:19:18 | anskiy | thank you! :) | |
| 12:19:27 | jm1 | gtema: it uses add_ips_to_server and passes a server object (with server details but without add_server_interfaces being called before) and auto_ips=True. https://review.opendev.org/c/openstack/ansible-collections-openstack/+/852119/8/plugins/modules/server.py#1093 | |
| 12:21:31 | jm1 | gtema: the issue seems to be that add_ips_to_server returns None instead of a server object | |
| 12:35:59 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: Replaced expensive get_server() and fixed issues in server module https://review.opendev.org/c/openstack/ansible-collections-openstack/+/852119 | |
| 12:36:58 | jm1 | gtema: oh... maybe the issue is obscured by the until keyword in ci test. i am trying something else and will ping you when i need more help | |
| 12:40:15 | gtema | hmm, ok. Such things we would need to be able to reproduce on bare devstack as well, cause only this will be useful | |
| 12:48:19 | jm1 | gtema: i have another issue with this add_ips_to_server function. this is what i want to accomplish: https://review.opendev.org/c/openstack/ansible-collections-openstack/+/828613/11/ci/roles/floating_ip/tasks/main.yml#271 | |
| 12:48:58 | jm1 | gtema: here it is called: https://review.opendev.org/c/openstack/ansible-collections-openstack/+/828613/11/plugins/modules/floating_ip.py#311 | |
| 12:49:38 | jm1 | gtema: this call returns None as one can see at timestamp "2022-08-10 09:47:26.103680" at https://e045d1163dd923a48c48-5f55b49caeaffd6ebf7a83fc93de4834.ssl.cf2.rackcdn.com/828613/11/check/ansible-collections-openstack-functional-devstack-ansible-2.12/218c3fb/job-output.txt | |
| 12:50:54 | jm1 | gtema: sorry, not the call to the function. the module returns null for floating_ip which means our code cannot find a floating ip | |
| 12:51:35 | jm1 | ..when searching for all floating ips attached to that server | |
| 12:52:23 | jm1 | gtema: i think add_ips_to_server somehow decides that it does not have to do anything. | |
| 12:53:11 | gtema | ;-( this is so hard to analyse. Best would be to have a server and to be able to try out function manually. Cause add_ips_to_server is an entry to the hell | |
| 12:53:38 | jm1 | gtema: sometimes it takes a while until a floating ip is listed for a server. but i handle that by waiting (with until) for the floating ip to be attached to the server... which never happens | |
| 12:54:02 | jm1 | gtema: actually this works in my local devstack but it fails in our ci.. | |
| 12:54:46 | gtema | he, and this is exactly what I told you yesterday: no single cloud (instead of devstack) is openstack cloud (tm) | |
| 12:55:11 | gtema | do we have here ip_pool or ips set? | |
| 12:55:29 | gtema | I guess only ip_pool | |
| 12:56:43 | gtema | madness is in reality in this case due to pool and attempt to reuse eventually existing fips | |
| 12:59:02 | gtema | for the sake of logs and reprodibility: can you add debug of the full server info and list all existing fips before this task is executed? | |
| 12:59:34 | gtema | maybe actually we can also try to start using sdk_logging here, but that would be simply huge | |
| 13:02:37 | opendevreview | ribaudr proposed openstack/python-openstackclient master: Microversion 2.91: Support specifying destination host to unshelve https://review.opendev.org/c/openstack/python-openstackclient/+/831902 | |
| 13:04:22 | opendevreview | ribaudr proposed openstack/python-openstackclient master: Microversion 2.91: Support specifying destination host to unshelve https://review.opendev.org/c/openstack/python-openstackclient/+/831902 | |
| 13:04:38 | jm1 | gtema: ack, will add debug statements | |
| 13:06:32 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: Refactored floating_ip and floating_ip_info modules https://review.opendev.org/c/openstack/ansible-collections-openstack/+/828613 | |
| 13:11:37 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: Refactored floating_ip and floating_ip_info modules https://review.opendev.org/c/openstack/ansible-collections-openstack/+/828613 | |
| 13:14:36 | jm1 | gtema: we could replace add_ips_to_server in aoc modules, but that would replicate code. this belongs to sdk i would think | |
| 13:15:15 | gtema | definitely. I want to gather few examples (conditions) when it fails and extend functional tests | |
| 14:07:15 | opendevreview | Arx Cruz proposed openstack/ansible-collections-openstack master: Update federation_mapping_info module to use proxy https://review.opendev.org/c/openstack/ansible-collections-openstack/+/849283 | |
| 14:46:56 | opendevreview | Merged openstack/openstacksdk master: Replace base64.encodestring with encodebytes https://review.opendev.org/c/openstack/openstacksdk/+/853940 | |
| 14:53:11 | jm1 | gtema: new log for first issue https://e3e2390f493f14093eaf-5e1de375231dae3d22db48edd8691108.ssl.cf5.rackcdn.com/852119/9/check/ansible-collections-openstack-functional-devstack-ansible-2.9/b2d8736/job-output.txt | |
| 14:54:37 | jm1 | gtema: add_ips_to_server with auto_ip=True should simply add a floating ip to server but fails with "No port on server XXX was found matching your NAT destination network XXX" | |
| 14:57:51 | jm1 | gtema: i guess you want some extra output. will add debug statements | |
| 14:59:30 | gtema | ok, thanks | |
| 14:59:37 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: Replaced expensive get_server() and fixed issues in server module https://review.opendev.org/c/openstack/ansible-collections-openstack/+/852119 | |
| 15:32:11 | jm1 | gtema: for the second issue with add_ips_to_server we have some debug output now https://dbb517edecb69c839e64-a1eedb38eefc88bab97a9b5aab34511e.ssl.cf1.rackcdn.com/828613/13/check/ansible-collections-openstack-functional-devstack-ansible-2.12/2bd2d7b/job-output.txt | |
| 15:32:42 | jm1 | gtema: this is the patch https://review.opendev.org/c/openstack/ansible-collections-openstack/+/828613 | |
| 15:33:33 | gtema | cool, thks | |
| 15:34:40 | gtema | okay, this means clearly we are in the "reuse existing unbound fip" | |
| 15:34:57 | gtema | this is what I was suspecting. | |
| 15:42:21 | gtema | @jm1: do you see that port_id in the query is of a fixed ip and not floating? | |
| 15:44:37 | jm1 | gtema:? | |
| 15:46:51 | gtema | "Get floating ip attached to server 2" task in the log uses "port": "1e08abff-72cd-465c-b92d-18806df814d6" as filter | |
| 15:47:09 | gtema | and this port is of a fixed address and not floating | |
| 15:48:16 | gtema | I am wondered that in the test you simply use server2_ports.ports[1].id | |
| 16:04:37 | opendevreview | mbu proposed openstack/osc-lib stable/train: Do not output log on bash autocomplete generation https://review.opendev.org/c/openstack/osc-lib/+/854016 | |
| 16:07:06 | matbu | Hey folks, is it possible to merge this patch, which is a cherry pick from master: https://review.opendev.org/c/openstack/osc-lib/+/854016 | |
| 16:07:10 | matbu | thank you by advance | |
| 16:07:25 | matbu | (and all the previous stack for wallaby & cie) | |
| 16:09:29 | opendevreview | Merged openstack/ansible-collections-openstack master: Deprecate special value 'auto' for attribute 'id' of compute_flavor https://review.opendev.org/c/openstack/ansible-collections-openstack/+/852936 | |
| 16:28:25 | opendevreview | Sharon Koech proposed openstack/python-openstackclient master: Add ``--cascade`` CLI option to network delete command https://review.opendev.org/c/openstack/python-openstackclient/+/854032 | |
| 16:44:02 | frickler | matbu: do you have info how to reproduce the issue to be fixed by that patch? to me the fix looks at least incomplete and shouldn't have been merged in master either | |
| 16:57:04 | matbu | frickler: I have a bug here: https://bugzilla.redhat.com/show_bug.cgi?id=2083742 | |
| 16:57:26 | matbu | frickler: i dont have much more information than that | |
| 16:57:48 | matbu | frickler: I was just tracking it, but I take a closer look to get a proper fix if needed | |
| 17:03:27 | frickler | matbu: thx, I just commented on the reviews in parallel. "Steps to reproduce" isn't really helpful, with that input I'd reject a bug as invalid. are you with the RDO team? getting a useable reproducer would be really helpful | |
| 17:04:01 | matbu | frickler: yep I will ask for a reproducer to the bug reporter | |
| #openstack-sdks - 2022-08-23 | |||
| 03:30:29 | opendevreview | yangjianfeng proposed openstack/python-openstackclient master: Add router ndp proxy commands https://review.opendev.org/c/openstack/python-openstackclient/+/749038 | |
| 03:37:24 | opendevreview | yangjianfeng proposed openstack/python-openstackclient master: Add router ndp proxy commands https://review.opendev.org/c/openstack/python-openstackclient/+/749038 | |
| 06:26:25 | opendevreview | yangjianfeng proposed openstack/python-openstackclient master: Add router ndp proxy commands https://review.opendev.org/c/openstack/python-openstackclient/+/749038 | |
| 07:27:15 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: [DNM] Refactored baremetal_node and baremetal_node_info modules https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854113 | |
| 07:53:09 | opendevreview | Sharon Koech proposed openstack/python-openstackclient master: Add ``--cascade`` CLI option to network delete command https://review.opendev.org/c/openstack/python-openstackclient/+/854032 | |
| 08:15:12 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: [DNM] Refactored baremetal_node and baremetal_node_info modules https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854113 | |
| 08:15:48 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: [DNM] Refactored baremetal_node and baremetal_node_info modules https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854113 | |
| 09:04:31 | opendevreview | ribaudr proposed openstack/openstacksdk master: Microversion 2.91: Support specifying destination host to unshelve https://review.opendev.org/c/openstack/openstacksdk/+/853949 | |
| 09:16:43 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: [DNM] Refactored baremetal_node and baremetal_node_info modules https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854113 | |
| 09:26:46 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: [DNM] Refactored baremetal_node and baremetal_node_info modules https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854113 | |
| 09:32:03 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: [DNM] Refactored baremetal_node and baremetal_node_info modules https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854113 | |
| 11:00:40 | opendevreview | Will Szumski proposed openstack/ansible-collections-openstack master: Allow networks to be updated https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854170 | |
| 11:22:37 | jm1 | gtema: can we rewrite register_machine to use baremetal/node resources? https://opendev.org/openstack/openstacksdk/src/commit/6bc56b0eb6eeb21569aae72ee8033ac64db54fb9/openstack/cloud/_baremetal.py#L160 | |
| 11:22:38 | jm1 | for example, we could fill a Node object with all kwargs and then use baremetal.create_node() instead of self._baremetal_client.post() | |
| 11:23:10 | jm1 | gtema: or will _baremetal_client.post() do exactly that? | |
| 11:28:36 | opendevreview | Will Szumski proposed openstack/ansible-collections-openstack master: Allow networks to be updated https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854170 | |
| 11:41:48 | opendevreview | Will Szumski proposed openstack/ansible-collections-openstack master: Allow networks to be updated https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854170 | |
| 11:44:09 | opendevreview | Will Szumski proposed openstack/ansible-collections-openstack master: Allow networks to be updated https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854170 | |
| 11:46:04 | opendevreview | Will Szumski proposed openstack/ansible-collections-openstack master: Allow networks to be updated https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854170 | |
| 11:53:05 | frickler | gtema: stephenfin: could you please add https://review.opendev.org/c/openstack/python-openstackclient/+/749038 to your review list? I added some questions for other reviewers in my comments. I know it is late in the cycle but it would be really great if this patch would make it for Zed. otherwise the feature is in Neutron but noone can use it. | |
| 11:59:59 | opendevreview | Areg Grigoryan proposed openstack/openstacksdk master: Bugfix for paginating for complex paths https://review.opendev.org/c/openstack/openstacksdk/+/853656 | |
| 12:37:34 | gtema | <jm1> "gtema: or will _baremetal_client..." <- it does exactly that | |
| 12:38:05 | gtema | frickler, but with this amount of comments ... Do you think they would be addressed quickly? | |
| 12:48:41 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: [DNM] Refactored baremetal_node and baremetal_node_info modules https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854113 | |
| 12:52:08 | jm1 | gtema: oh cool! so this should work? https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854113/7/plugins/modules/baremetal_node.py#655 | |
| 12:53:15 | jm1 | gtema: kwargs uses attribute names of the Node object such as 'chassis_id' and no longer ironic names such as 'chassis_uuid' | |
| 12:54:29 | opendevreview | Jakob Meng proposed openstack/ansible-collections-openstack master: Refactored baremetal_node and baremetal_node_info modules https://review.opendev.org/c/openstack/ansible-collections-openstack/+/854113 | |
| 12:59:50 | gtema | jm1: for BM I would really suggest to try not to use cloud layer. Because I have no clue whether and how ironic folks use it I have fear to touch it really. whatever is on the proxy side is always verified by tests and therefore can be fixed when necessary | |
| 13:07:36 | jm1 | gtema: but that would imply replicating code from the sdk in aoc | |
| 13:08:19 | gtema | which particular cases? | |
| 13:08:36 | jm1 | gtema: maybe we should ask dtantsur how or if ironic is using this register_machine function.. | |
| 13:08:47 | gtema | I was meaning similarly to the create_server case: for the complex cases of course we use cloud. | |
| 13:09:08 | dtantsur | I highly suspect we do use it | |
| 13:09:28 | jm1 | gtema: register_machine is not only creating a node but also creating baremetal nodes | |
| 13:09:46 | frickler | gtema: yes, I've been in contact with the author already | |
| 13:11:24 | frickler | I just would like to avoid needing another set of iterations when other reviewers have different views than me | |
| 13:11:35 | jm1 | dtantsur: codesearch does not return any match. how could ironic possibly use it? maybe though aoc? https://codesearch.opendev.org/?q=register_machine&i=nope&literal=nope&files=&excludeFiles=&repos= | |