| *** MickyMan77 has joined #openstack-ansible | 00:36 | |
| *** MickyMan77 has quit IRC | 00:45 | |
| *** MickyMan77 has joined #openstack-ansible | 01:20 | |
| *** mpsairam has quit IRC | 01:27 | |
| *** pmannidi has joined #openstack-ansible | 01:27 | |
| *** MickyMan77 has quit IRC | 01:29 | |
| *** djhankb has quit IRC | 01:29 | |
| *** djhankb has joined #openstack-ansible | 01:30 | |
| *** rfolco has joined #openstack-ansible | 01:48 | |
| *** rfolco has quit IRC | 02:29 | |
| *** pmannidi has quit IRC | 03:03 | |
| *** maharg101 has joined #openstack-ansible | 03:17 | |
| *** maharg101 has quit IRC | 03:22 | |
| *** MickyMan77 has joined #openstack-ansible | 03:29 | |
| *** MickyMan77 has quit IRC | 03:38 | |
| *** maharg101 has joined #openstack-ansible | 03:54 | |
| *** maharg101 has quit IRC | 03:59 | |
| *** MickyMan77 has joined #openstack-ansible | 04:00 | |
| *** pmannidi has joined #openstack-ansible | 04:19 | |
| *** evrardjp has quit IRC | 04:33 | |
| *** evrardjp has joined #openstack-ansible | 04:33 | |
| *** djhankb has quit IRC | 04:45 | |
| *** djhankb has joined #openstack-ansible | 04:46 | |
| *** MickyMan77 has quit IRC | 05:04 | |
| *** maharg101 has joined #openstack-ansible | 05:55 | |
| *** dmsimard has quit IRC | 05:55 | |
| *** dmsimard has joined #openstack-ansible | 05:58 | |
| *** MickyMan77 has joined #openstack-ansible | 06:00 | |
| *** maharg101 has quit IRC | 06:01 | |
| *** MickyMan77 has quit IRC | 06:09 | |
| *** miloa has joined #openstack-ansible | 06:16 | |
| noonedeadpunk | mornings | 06:37 |
|---|---|---|
| *** MickyMan77 has joined #openstack-ansible | 06:45 | |
| *** viks____ has joined #openstack-ansible | 06:47 | |
| ebbex | mornin' | 06:51 |
| ebbex | noonedeadpunk: I managed to get the ceph cluster issue again. | 06:52 |
| ebbex | INSERT INTO `clusters` VALUES ('2020-10-10 19:48:16','2020-10-10 19:48:16',NULL,0,2,'ceph@RBD','cinder-volume',0,NULL,0,'disabled',NULL,0); | 06:53 |
| *** MickyMan77 has quit IRC | 06:54 | |
| ebbex | And was created by having just one cinder-volume.service start with cinder.conf containing [RBD]\nbackend_host: rbd:volumes | 06:54 |
| recyclehero | mornings guys | 06:55 |
| recyclehero | Finally satisfied with my test setup. I wrote a script to convert all the vm hard drives from the failed queens enviroment to qcow2, add a public key to all of them for easy ssh auth, and upload them to my new os glance. | 06:57 |
| recyclehero | apart from getting some 504s on horizions it seems ok | 06:57 |
| recyclehero | still no cors for irect glance upload | 06:57 |
| recyclehero | noonedeadpunk: there is some issues needs exprt attention | 06:58 |
| noonedeadpunk | ebbex: huh, so it doesn't really matter if we have backend_host defined or not | 06:58 |
| recyclehero | compute1 was getting an ipv6 automatically that prevented it from creating bridges | 06:58 |
| noonedeadpunk | and this was the request with which it failed, or which is needed to fix? | 06:58 |
| recyclehero | compute1 AppArmor has some issues with NFS so I had to disable it | 06:59 |
| *** andrewbonney has joined #openstack-ansible | 07:00 | |
| *** tosky has joined #openstack-ansible | 07:03 | |
| * noonedeadpunk needs to spawn aio | 07:04 | |
| ebbex | noonedeadpunk: yup. for now it seems only that the create statement is sometimes missing, and none of the ceph-volume services will create it on startup if backend_host is not set. | 07:04 |
| noonedeadpunk | I'm wondering if we're missing some command to run, (ie cinder-manage one) | 07:05 |
| ebbex | but i did deploy once without issues, only thing different was i had 'cinder_ceph_client: cinder' and 'glance_ceph_client: glance' in my user_variables.yml | 07:06 |
| noonedeadpunk | ebbex: I guess it's default? https://opendev.org/openstack/openstack-ansible/src/branch/master/inventory/group_vars/all/cinder.yml#L17 | 07:08 |
| ebbex | Yeah, i figured they were the defaults so i removed them, and the two times i've deployed without them, something went wrong. | 07:11 |
| noonedeadpunk | totally `wtf` sotuation.... | 07:12 |
| noonedeadpunk | *situation | 07:12 |
| ebbex | yup. | 07:12 |
| *** rpittau|afk is now known as rpittau | 07:35 | |
| jrosser | ebbex: we do have some odd sequencing perhaps? | 07:44 |
| *** maharg101 has joined #openstack-ansible | 07:45 | |
| jrosser | like install the services, drop the systemd units, adjust the conf files with the templates, some restarts.... | 07:45 |
| jrosser | anyway, good morning all | 07:49 |
| *** sshnaidm|off is now known as sshnaidm | 07:58 | |
| noonedeadpunk | I think I have some time during upcoming weeks to look into that, just not quite understand how to reproduce that... just drop glance_ceph_client and cinder_ceph_client from user variables and leave only in group vars... | 08:02 |
| noonedeadpunk | what is clear, is that aio in CI does not reproduce issue... | 08:11 |
| jrosser | recyclehero: for your NFS/apparmor problem when mouting into a container, or onto the host? | 08:15 |
| *** MickyMan77 has joined #openstack-ansible | 08:35 | |
| MickyMan77 | jrosser: I have manually link the libs but when I try to start the services I get this error... "TabError: inconsistent use of tabs and spaces in indentation". -- http://paste.openstack.org/show/798938/ | 08:37 |
| noonedeadpunk | gerritbot seems broken again | 08:55 |
| jrosser | MickyMan77: you shouldnt need to manually link the libraries, the ceph_client role should do that for you i think | 09:05 |
| recyclehero | jrosser: into the host. I set up storage backend as nfs so it tries to mount also in the compute physical host. | 09:13 |
| jrosser | recyclehero: well i guess that is a debian / apparmor specific thing, if you know what we should do to allow that nfs mount then it might be possible to patch it, or put in the docs what must be done for preparing the hosts | 09:16 |
| *** openstackgerrit has joined #openstack-ansible | 09:16 | |
| openstackgerrit | wu.shiming proposed openstack/openstack-ansible master: Use unittest.mock instead of mock https://review.opendev.org/757490 | 09:16 |
| *** djhankb has quit IRC | 09:18 | |
| *** djhankb has joined #openstack-ansible | 09:19 | |
| MickyMan77 | jrosser: If I stand in venvs/cinder*** folder and do an ". bin/activate" and then I run "bin/pip3 install python-cephlibs" that will install the python files. | 09:24 |
| openstackgerrit | Merged openstack/openstack-ansible-os_nova master: Trigger uwsgi restart https://review.opendev.org/756861 | 09:24 |
| jrosser | the ceph python libs need to be installed on the host and then symlinked into the venv | 09:25 |
| recyclehero | jrosser: sure, I will share what I have done about it. but maybe the problem was the bridge problem which also need addressing. the interface for flat-vlan network which is passed to compute node (host_override_interface) was getting an IPV6 address which was cauging lot of problems | 09:25 |
| jrosser | this is because the python libs are a thin shim around librbd and if the python bindings do not exactly match the shared libraries on the host then it's all going to go badly | 09:25 |
| jrosser | MickyMan77: i would really recommend using only the playbooks to set all this up, not manual installation because it's doing some really quite specific things | 09:26 |
| recyclehero | why are the oversubscription of nova was chosen this much lower than the OS defaults? | 09:27 |
| recyclehero | OS nova_cpu_allocation_ratio:16 OSA nova_cpu_allocation_ratio:2 | 09:27 |
| jrosser | in fact more specifically the ceph python libs need to be installed on the host from the apt packages from the ceph site to be matching with the librbd packages, not using pip | 09:27 |
| jrosser | recyclehero: it depends on your workload, override it to whatever you want | 09:28 |
| openstackgerrit | Merged openstack/openstack-ansible-os_glance master: Trigger uwsgi restart https://review.opendev.org/756681 | 09:28 |
| ebbex | jrosser: maybe, i don't really see how. I'll try a couple redeploys tonight to see if i can pin it down to a specific combination. | 09:30 |
| MickyMan77 | jrosser: can I just run the playbook ceph_install_python_libs.yml ? | 09:45 |
| jrosser | no, thats a set of tasks in an ansible role | 09:45 |
| *** MickyMan77 has quit IRC | 09:46 | |
| jrosser | the ceph_client role is run from the os_cinder role, if the appropriate condition is met, see https://github.com/openstack/openstack-ansible-os_cinder/blob/master/tasks/main.yml#L166-L176 | 09:48 |
| openstackgerrit | Merged openstack/openstack-ansible-os_aodh master: Trigger uwsgi restart https://review.opendev.org/756684 | 09:48 |
| openstackgerrit | Merged openstack/openstack-ansible master: Switch integrated linters to focal https://review.opendev.org/755759 | 09:59 |
| openstackgerrit | Merged openstack/openstack-ansible-ceph_client master: Fix linter errors https://review.opendev.org/755697 | 09:59 |
| openstackgerrit | Merged openstack/openstack-ansible-ceph_client master: Add configuration example for Ceph configuration from files. https://review.opendev.org/756051 | 09:59 |
| openstackgerrit | Merged openstack/openstack-ansible-ceph_client master: Allow to proceed with role if ceph_conf_file is set https://review.opendev.org/755820 | 09:59 |
| openstackgerrit | wu.shiming proposed openstack/openstack-ansible-os_cinder master: Dep's should be restricted by upper-constraints https://review.opendev.org/757525 | 10:01 |
| *** shyamb has joined #openstack-ansible | 10:17 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_horizon master: Updated from OpenStack Ansible Tests https://review.opendev.org/754169 | 10:18 |
| openstackgerrit | likui proposed openstack/openstack-ansible master: Reuse the docs deps to benefit from constraints https://review.opendev.org/757443 | 10:20 |
| openstackgerrit | Merged openstack/openstack-ansible master: Remove "when" statement from vars_prompt https://review.opendev.org/755824 | 10:28 |
| *** shyamb has quit IRC | 10:31 | |
| *** pcaruana has quit IRC | 10:31 | |
| recyclehero | regarding overcommiting memory, is ksm avaialble by default for nova kvm driver? | 10:37 |
| *** djhankb has quit IRC | 10:38 | |
| *** djhankb has joined #openstack-ansible | 10:38 | |
| *** shyamb has joined #openstack-ansible | 10:58 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_horizon master: Do not manage /etc/ssl or /etc/pki directories or symlinks https://review.opendev.org/754093 | 10:59 |
| *** pcaruana has joined #openstack-ansible | 11:03 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Add infra testing scenario https://review.opendev.org/755497 | 11:14 |
| jrosser | recyclehero: you can enable KSM is you need it https://github.com/openstack/openstack-ansible-os_nova/blob/master/defaults/main.yml#L170 | 11:14 |
| jrosser | noonedeadpunk: SHA bump on master makes unfortunate error for debian https://zuul.opendev.org/t/openstack/build/4ac0f9f137fe4522abbd4b1dcc8ebcca/log/logs/host/nova-compute.service.journal-10-34-13.log.txt#4516-4527 | 11:17 |
| noonedeadpunk | jrosser: uh, seems we need to enable buster-backports for qemu | 11:18 |
| noonedeadpunk | as its 3.1 vs 5,0 | 11:19 |
| jrosser | makes me wonder why that has broken so late | 11:22 |
| jrosser | you would change that sort of thing right at the start of a cycle rather than just at the end? | 11:22 |
| noonedeadpunk | I hope I bumped really V there | 11:23 |
| *** shyamb has quit IRC | 11:24 | |
| *** shyamb has joined #openstack-ansible | 11:24 | |
| noonedeadpunk | well yes, it's just landed https://opendev.org/openstack/nova/commit/2931516b75ab1eda8deefec6910efee82a0a7ae2 | 11:24 |
| *** shyam89 has joined #openstack-ansible | 11:24 | |
| noonedeadpunk | uh | 11:24 |
| *** shyam89 has quit IRC | 11:25 | |
| *** MickyMan77 has joined #openstack-ansible | 11:34 | |
| *** rh-jelabarre has joined #openstack-ansible | 11:54 | |
| *** shyamb has quit IRC | 11:57 | |
| *** shyam89 has joined #openstack-ansible | 11:57 | |
| *** recyclehero has quit IRC | 11:57 | |
| *** recyclehero has joined #openstack-ansible | 12:03 | |
| openstackgerrit | Merged openstack/openstack-ansible master: Simplify path for letsencrypt usage https://review.opendev.org/751327 | 12:05 |
| *** shyam89 has quit IRC | 12:07 | |
| *** shyam89 has joined #openstack-ansible | 12:08 | |
| *** shyamb has joined #openstack-ansible | 12:19 | |
| *** shyam89 has quit IRC | 12:20 | |
| *** shyamb has quit IRC | 12:26 | |
| *** soutr has joined #openstack-ansible | 12:30 | |
| jrosser | noonedeadpunk: i think we have something a bit broken in our strategy plugin, see this https://zuul.opendev.org/t/openstack/build/419d108ff609461f95300206dd5c9262/log/job-output.txt#15345-15346 | 12:47 |
| *** yann-kaelig has joined #openstack-ansible | 12:47 | |
| jrosser | the task returns no status at all, so not skipped / ok / failed, just nothing | 12:47 |
| jrosser | and that also means that the task is not in the ARA report by the look of it | 12:47 |
| jrosser | we have an early exit point here https://github.com/openstack/openstack-ansible-plugins/blob/master/strategy/linear.py#L124-L125 | 12:48 |
| noonedeadpunk | jrosser: well, that never caused any actual issues | 12:49 |
| jrosser | that checks any conditionals on the task | 12:49 |
| noonedeadpunk | and I think here issue is in patch itself... | 12:49 |
| noonedeadpunk | but yeah, I noticed that previously as well | 12:49 |
| jrosser | i guess i was surprised not to see the task at all in ARA | 12:49 |
| noonedeadpunk | well, ara changed the way of logging I think as well) | 12:50 |
| noonedeadpunk | from other side because of early exit we probably save up some time? | 12:50 |
| jrosser | thats the idea i suppose, though it is from very long ago | 12:50 |
| noonedeadpunk | yeah it is.... | 12:50 |
| noonedeadpunk | well yes, it's probably worth looking at one day | 12:51 |
| noonedeadpunk | jrosser: btw offtopic question - I guess you're using barbican? | 12:53 |
| jrosser | i'm just taking a look at keystone why the uw_apache test seems to fail with missing rsync | 12:53 |
| jrosser | seems related to cleaning up the lxc packages by i can't see what | 12:54 |
| jrosser | noonedeadpunk: no we don't use barbican, but it would be nice | 12:54 |
| jrosser | no good answer to the secrets backend is kind of my issue with it | 12:54 |
| noonedeadpunk | well ye felt it:) | 12:54 |
| jrosser | probably hashi vault would be my first choice | 12:55 |
| noonedeadpunk | actually I was trying to use vault one | 12:55 |
| noonedeadpunk | yeah. and it seems there's no need in barbican to use vault... | 12:55 |
| jrosser | we do use vault, just not for this yet, so have some experience with it | 12:55 |
| redrobot | 👀 | 12:56 |
| noonedeadpunk | moreover, barbican weirdly fails when it set to store with vault and cinder tries to read previously created secret | 12:56 |
| noonedeadpunk | redrobot: I tried to write in #openstack-barbican :p | 12:56 |
| noonedeadpunk | redrobot: http://paste.openstack.org/show/798944/ | 12:56 |
| noonedeadpunk | what is weird, that when I just manually put data it's readed fine | 12:57 |
| redrobot | Oh weird | 12:57 |
| noonedeadpunk | http://paste.openstack.org/show/798948/ | 12:57 |
| noonedeadpunk | when cinder is set to use vault directly it's ok as well.... | 12:58 |
| noonedeadpunk | (directly through castellian I mean) | 12:58 |
| redrobot | So Barbican with Vault backend is borked, but Castellan with Vault backend is OK? | 12:59 |
| redrobot | 🤔 | 12:59 |
| noonedeadpunk | kind of.... | 12:59 |
| noonedeadpunk | which does not make much sense seing https://opendev.org/openstack/barbican/src/branch/master/barbican/plugin/vault_secret_store.py | 13:00 |
| noonedeadpunk | redrobot: well, barbican with vault is ok, until cinder tries to read previously created payload | 13:01 |
| noonedeadpunk | but, what makes me suspicios, is metadata for cinder created key.... | 13:02 |
| noonedeadpunk | volume type is `openstack volume type create --encryption-provider luks --encryption-cipher aes-xts-plain64 --encryption-key-size 256 --encryption-control-location front-end LUKS --property volume_backend_name='rbddriver'` | 13:02 |
| noonedeadpunk | but it results in ` Mode | None ` for the secret | 13:02 |
| redrobot | yeah, mode is just metadata, it doesn't really do much. | 13:03 |
| redrobot | > previously created payload < Previously created how? 🤔 | 13:03 |
| noonedeadpunk | I mean just created volume | 13:03 |
| noonedeadpunk | as api/scheduler part is passed and it fails on cinder-volume part | 13:04 |
| noonedeadpunk | So I guess that payload should be already created and stored? | 13:04 |
| noonedeadpunk | (by cinder) | 13:04 |
| noonedeadpunk | redrobot: I have actually a sandbox where I'm playing with it... but have no idea what to look for at the moment | 13:11 |
| noonedeadpunk | redrobot: oh, well, and `encoded` value which is passed to oslo_serialization.base64 is `b'\xe2\xca7\xc2i$\x99\x8a\xb5\xf6r\x9d&@\xe6\xf1\xfbW\xfeL\x1d\xae\xa1\x8e\x1b\xe6\xea\xaa\xc0\xf5\xd3\xa1'` which I have no idea how to decrypt as well | 13:14 |
| noonedeadpunk | s/decrypt/decode | 13:14 |
| noonedeadpunk | so actually what results in this error is `openstack secret get https://10.6.0.196:9311/v1/secrets/ed50129c-eb0e-4fe3-8e43-696197854768 -t application/octet-stream -p` | 13:21 |
| *** maharg101 has quit IRC | 13:22 | |
| noonedeadpunk | so my guess is that maybe it's not properly decrypted.... | 13:28 |
| *** soutr has quit IRC | 13:30 | |
| recyclehero | if I want to get rid of neutron-metering-agent Should I override it and remove it from contains list of network-containers via a override in /etc/.../env.d or just remove it from neutron plugins in its conf? | 13:32 |
| recyclehero | or both? | 13:32 |
| *** pcaruana has quit IRC | 13:35 | |
| noonedeadpunk | redrobot: hm, just out of interest I placed "raise ValueError" here https://opendev.org/openstack/barbican/src/branch/master/barbican/plugin/resources.py#L71 - and it's not raised, but only http://paste.openstack.org/show/798944/ | 13:42 |
| noonedeadpunk | so it seems I was wrong about that it has stored somewhere | 13:42 |
| noonedeadpunk | but how/why it tries to read data then.... | 13:42 |
| noonedeadpunk | it's stack trace from cinder size... http://paste.openstack.org/show/798953/ | 13:43 |
| *** d34dh0r53 has quit IRC | 13:54 | |
| *** d34dh0r53 has joined #openstack-ansible | 14:00 | |
| *** gshippey has joined #openstack-ansible | 14:10 | |
| ebbex | http://paste.openstack.org/show/798955/ is this the correct format for pci passthru ? | 14:18 |
| ebbex | ah, not quite. nova_pci_passthrough_whitelist is not actually a list :/ | 14:20 |
| jrosser | i had something like this http://paste.openstack.org/show/798956/ | 14:22 |
| ebbex | jrosser: i started out using nova_nova_conf_overrides, then I saw in the template these other options. Thanks :) | 14:43 |
| *** MickyMan77 has quit IRC | 15:12 | |
| *** miloa has quit IRC | 15:23 | |
| jrosser | ooohhh haproxy/metal is passing https://review.opendev.org/#/c/733408/ | 15:35 |
| jrosser | the thing blocking that merging would be this https://review.opendev.org/#/c/754092/ | 15:36 |
| *** cyberpear has joined #openstack-ansible | 15:43 | |
| jrosser | i ran the os_keystone uw_apache functional test locally and it passed | 15:49 |
| jrosser | not sure what is going on there with not able to find 'rsync' | 15:49 |
| noonedeadpunk | centos 8 looks the way worse. I thought we even disabled it? or it was for some other role? | 15:50 |
| jrosser | hmm? iirc theres is a seperate patch to add centos-8 there? | 15:51 |
| noonedeadpunk | well, rsync added for redhat but not for uubntu/debian images with https://review.opendev.org/#/c/756587 | 15:52 |
| jrosser | keystone role installs rsync too | 15:52 |
| jrosser | this is whats so confusing | 15:52 |
| noonedeadpunk | well yeah, it does... | 15:52 |
| noonedeadpunk | oh wait | 15:53 |
| noonedeadpunk | it's missing rsync on the host where it tries to connect | 15:53 |
| noonedeadpunk | which is 10.1.0.4 | 15:53 |
| noonedeadpunk | and it's keystone2? | 15:54 |
| jrosser | hmm - it's not running serial is it | 15:54 |
| jrosser | and the package installation has not yet happened on the 2nd instance..... | 15:55 |
| noonedeadpunk | it is https://opendev.org/openstack/openstack-ansible-tests/src/branch/master/test-install-keystone.yml#L27 | 15:55 |
| jrosser | matching here https://github.com/openstack/openstack-ansible/blob/master/playbooks/os-keystone-install.yml#L39 | 15:56 |
| noonedeadpunk | yeah.... | 15:57 |
| noonedeadpunk | so we kind of need rsync before role get's launched.... | 15:57 |
| jrosser | unfortunate | 15:57 |
| noonedeadpunk | we probably ned to leave comments _why_ we need packages in lxc_hosts | 15:58 |
| jrosser | agreed | 15:58 |
| jrosser | theres also this comment too https://github.com/openstack/openstack-ansible-os_keystone/blob/4c9dda7dc4e90ade994eb05928a2d636c6747f43/tasks/keystone_fernet_keys_distribute.yml#L16 | 16:00 |
| jrosser | but iirc synchronize module needs rsync anyway | 16:00 |
| noonedeadpunk | yes, it is | 16:00 |
| *** rpittau is now known as rpittau|afk | 16:01 | |
| noonedeadpunk | well, it will be more neat, but yeah, we need tsync anyway) | 16:01 |
| noonedeadpunk | *rsync | 16:01 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-lxc_hosts master: Add rsync to container base image https://review.opendev.org/757612 | 16:11 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-os_keystone master: Do not manage /etc/ssl or /etc/pki directories or symlinks https://review.opendev.org/754092 | 16:11 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-os_cloudkitty master: Define condition for the first play host one time https://review.opendev.org/751261 | 16:20 |
| *** tosky has quit IRC | 16:21 | |
| jrosser | i think this is stopping things merge in python_venv_build role https://review.opendev.org/#/c/756279/ | 16:22 |
| *** spatel has joined #openstack-ansible | 16:28 | |
| spatel | I know its too early but do we know when openstack-ansible victoria stable going to come out? | 16:30 |
| spatel | I am planning to skip Ussuri if victoria is current stable or close to stable | 16:30 |
| *** csmart has quit IRC | 16:40 | |
| jrosser | ebbex: http://paste.openstack.org/show/798963/ | 17:07 |
| *** pcaruana has joined #openstack-ansible | 17:13 | |
| *** andrewbonney has quit IRC | 17:29 | |
| *** gyee has joined #openstack-ansible | 18:00 | |
| ebbex | jrosser: well that's new... how does that apply to something? do you combine it to a dict, and it will set/cancel out a key with null? | 18:10 |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron master: Updated from OpenStack Ansible Tests https://review.opendev.org/755540 | 18:11 |
| jrosser | afaik it it a dict with keys but no values | 18:11 |
| jrosser | and it’s used several times in the systemd_networkd role examples | 18:11 |
| jrosser | would need a bit of trying out to see why it’s done that way I guess | 18:12 |
| *** spatel has quit IRC | 18:32 | |
| ebbex | jrosser: I'm guessing it's another way to make it look like a list, but behave like a dict? Though you could get the same result just adding a ':' at the end of the quoted key. | 18:35 |
| ebbex | I'm guessing cloudnull could chime in on that. | 18:37 |
| cloudnull | wat? | 18:37 |
| ebbex | cloudnull: Why are some "lists" using '?' instead of '-' in systemd_networkd | 18:38 |
| ebbex | ? | 18:38 |
| *** viks____ has quit IRC | 18:38 | |
| ebbex | cloudnull: like this, https://review.opendev.org/#/c/733408/19/tests/roles/bootstrap-host/tasks/prepare_networking.yml | 18:39 |
| jrosser | that syntax i took from the example here https://github.com/openstack/ansible-role-systemd_networkd/blob/master/examples/playbook.yml#L175-L180 | 18:40 |
| ebbex | curious side-question, what do you guys use for irc clients, and how do you get notifications on mentions? | 18:45 |
| cloudnull | ebbex that's a set vs an array | 18:46 |
| cloudnull | when using config_template we differentiate between the two data sources | 18:46 |
| cloudnull | there used to be docs on this , though im failing to find it | 18:48 |
| cloudnull | the TL;DR is , a set allows INI style config to create a duplicate entry within the section | 18:49 |
| cloudnull | openstack allows for multi-config-op which is not actually part of the ini spec. | 18:50 |
| cloudnull | to support that, config_template sees a set (items with ?) as a multi-config op when writing an INI file | 18:51 |
| cloudnull | when not writing an INI file , ? or - are just lists | 18:51 |
| cloudnull | https://github.com/openstack/ansible-config_template/blob/master/action/config_template.py#L136-L182 | 18:52 |
| cloudnull | that doc string is in code, I thought something more robust was published to, but I cant find it | 18:53 |
| cloudnull | ebbex I hope that answers your question | 18:53 |
| * cloudnull goes back to lurking | 18:53 | |
| *** djhankb has quit IRC | 18:54 | |
| jrosser | cloudnull: oh thats really neat | 18:54 |
| jrosser | we were looking at systemd timer units wanting multiple OnCalendar keys | 18:54 |
| *** djhankb has joined #openstack-ansible | 18:54 | |
| jrosser | that wouldallow it | 18:54 |
| ebbex | cloudnull: Thanks :) | 18:54 |
| cloudnull | ++ | 18:55 |
| cloudnull | jrosser yup that would allow it | 18:55 |
| cloudnull | we used it in openstack, mysql, and systemd config | 18:55 |
| *** yann-kaelig has quit IRC | 19:00 | |
| *** cshen has quit IRC | 19:01 | |
| jrosser | ebbex: i use irccloud, browser based. it hooks into desktop notifications from chrome, and also theres a mobile app | 19:02 |
| ebbex | I'll look into that, thanks :) | 19:15 |
| *** masterpe has quit IRC | 19:16 | |
| *** ioni has quit IRC | 19:16 | |
| *** fridtjof[m] has quit IRC | 19:16 | |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-os_nova master: Define condition for the first play host one time https://review.opendev.org/756332 | 19:25 |
| *** ioni has joined #openstack-ansible | 19:26 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_ceilometer master: Updated from OpenStack Ansible Tests https://review.opendev.org/755530 | 19:26 |
| openstackgerrit | Jonathan Rosser proposed openstack/ansible-role-python_venv_build master: Install git package on hosts building venvs https://review.opendev.org/756939 | 19:26 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-os_nova master: Simplify scheduler filter additions https://review.opendev.org/650144 | 19:29 |
| *** masterpe has joined #openstack-ansible | 19:50 | |
| *** fridtjof[m] has joined #openstack-ansible | 19:50 | |
| *** MickyMan77 has joined #openstack-ansible | 20:00 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_horizon master: Fix boolean in config file. https://review.opendev.org/755508 | 20:14 |
| *** djhankb has quit IRC | 20:15 | |
| *** djhankb has joined #openstack-ansible | 20:15 | |
| *** d34dh0r53 has quit IRC | 20:20 | |
| *** d34dh0r53 has joined #openstack-ansible | 20:25 | |
| *** tosky has joined #openstack-ansible | 20:26 | |
| recyclehero | 21.0.1: there is an option in neutron role, neutorn_plugin_base. I cant find an option with the same name in OS Neutron conf. is there the default options name for roles arbitrary? | 20:30 |
| *** d34dh0r53 has quit IRC | 20:30 | |
| recyclehero | if they are arbitrary chosen I think neutron_plugin_base maps to service_plugins in neutron.conf, right? | 20:30 |
| recyclehero | where can I see these mapping? | 20:31 |
| *** d34dh0r53 has joined #openstack-ansible | 20:31 | |
| *** d34dh0r53 has quit IRC | 20:35 | |
| *** d34dh0r53 has joined #openstack-ansible | 20:36 | |
| jrosser | recyclehero: unfortunately life is not always as simple as taking the config key name and exposing it directly as a variable. Sometimes extra logic is required, and that is the case for service_plugins in neutron.conf | 20:40 |
| *** yann-kaelig has joined #openstack-ansible | 20:41 | |
| jrosser | this is a useful way of finding out where all these things go http://codesearch.openstack.org/?q=neutron_plugin_base | 20:41 |
| *** yann-kaelig has quit IRC | 20:41 | |
| recyclehero | life wasnt easy this past week working with ansible, but its starting to payoff. thansk jrosser | 20:42 |
| jrosser | you'll see from that search that neutron_plugin_base is a term from inside the neturon code, so the choice is not arbitrary | 20:42 |
| recyclehero | jrosser: say I deployed neutron-metering-agent, then I go and delete it in my overridden env.d and also remove it from neutron_plugin_base. will it get removed? | 20:42 |
| jrosser | i don't know | 20:43 |
| jrosser | i would have to try | 20:43 |
| recyclehero | or see the playbooks to findout | 20:43 |
| jrosser | yes | 20:43 |
| recyclehero | but I think the playbooks are not so inclined to deleting things | 20:44 |
| jrosser | ansible in general is additive | 20:48 |
| jrosser | but you can declare state, like a service being started or stopped | 20:48 |
| recyclehero | in this case the neutorn-metering-agent is still there | 20:49 |
| MickyMan77 | Now I have another problem. I'm trying to deploy ceph-client. When I activate "ceph_keyrings_dir: /etc/openstack_deploy/ceph-keyrings" I do get this error --> http://paste.openstack.org/show/798966/ | 20:49 |
| jrosser | recyclehero: the way to approach this is to look in the os_neutron defaults/main.yml | 20:50 |
| jrosser | the defaults files for the roles are the "external interface" where settings are declared | 20:50 |
| jrosser | here would be the enable/disable flag for the neutron metring agent https://opendev.org/openstack/openstack-ansible-os_neutron/src/branch/master/vars/main.yml#L397 | 20:51 |
| jrosser | then we can find the condition which determines if that is true/false https://opendev.org/openstack/openstack-ansible-os_neutron/src/branch/master/vars/main.yml#L144 | 20:51 |
| recyclehero | jrosser: thanks I saw the deafualt and yet didnt see that | 20:52 |
| jrosser | so perhaps checking if 'metering' is in neutron_plugin_base is first step | 20:52 |
| jrosser | if so remove it, and the metering service will still be deployed, but hopefully the systemd unit will be disabled | 20:53 |
| recyclehero | but if I remove it from env.d neutron_agents_container maybe it shoudnt get deployed | 20:54 |
| jrosser | you can do that but i would not recommend that | 20:54 |
| recyclehero | care to elaborate? | 20:56 |
| recyclehero | jrosser: also I was looking at default/main.yml not vars/main.yml | 20:57 |
| jrosser | adjusting env.d and removing the definition of a group means that you then need to delete that group and all references to it from the dynamic inventory cache | 20:57 |
| *** cshen has joined #openstack-ansible | 20:58 | |
| jrosser | and perhaps also you'd want to check that the playbooks make no direct reference to the group you want to remove | 20:58 |
| jrosser | i.e they run without error if the group is undefined, rather than simply empty | 20:58 |
| recyclehero | I am playing now to find a near perfect deployemnt. if I dont have that from start the problem with inventory cache will be gone, right? | 20:59 |
| recyclehero | but I get it, its better to play with the options when u can and dont touch the skeleton as u mentioned earlier | 21:00 |
| *** cshen has quit IRC | 21:02 | |
| jrosser | in principle either way will work | 21:03 |
| jrosser | turning the metering agent on/off via neutron_plugin_base is quite simple and you could do that on a live deployment quite easily | 21:04 |
| openstackgerrit | Merged openstack/openstack-ansible-os_panko master: Cleanup apache tasks https://review.opendev.org/756864 | 21:04 |
| jrosser | asjusting env.d feels mich more invasive somehow | 21:04 |
| recyclehero | yes so there is no need doing anything with vars. but as I said it seems If you get it, you cant get rid of it. but I omitted it from neutron_service_plugins and that change hopeyfully projected to the nuetorn-server container. so I should be abte to safely disable the systemd for it or even delete its bin from /openstack | 21:06 |
| recyclehero | i removed it from base plugins but it still shows as running | 21:07 |
| recyclehero | jrosser: whats the purpose of component_skel in env.d files? I think they are internal and with dont see them on the surface | 21:12 |
| jrosser | https://opendev.org/openstack/openstack-ansible/src/branch/master/etc/openstack_deploy/env.d/extra_container.yml.example | 21:29 |
| jrosser | recyclehero: if the metering agent is still running i think you've now got enough info to trace through | 21:29 |
| jrosser | you can throw in a debug; task to print the value of neutron_metering | 21:30 |
| recyclehero | thats good idea and make me learn to write tasks | 21:33 |
| recyclehero | but it made me realize where my fault lays | 21:33 |
| recyclehero | in user_variables I wore neutorn_base_plugins: -router. so they would get merged from default and the final neutron_base_plugin would be -router and -metring. | 21:34 |
| jrosser | well actually it looks like a bug tbh https://opendev.org/openstack/openstack-ansible-os_neutron/src/branch/master/vars/main.yml#L325-L340 | 21:34 |
| recyclehero | I should do neutron_base_plugin: router(with out the dash) in user_variables? | 21:34 |
| jrosser | it's a yaml format list https://opendev.org/openstack/openstack-ansible-os_neutron/src/branch/master/defaults/main.yml#L338-L340 | 21:36 |
| jrosser | so one entry can be a newline with a dash | 21:36 |
| jrosser | or key: [ value ] | 21:36 |
| jrosser | as an alternate yaml form | 21:36 |
| jrosser | anyway late here, i'm done for today | 21:37 |
| recyclehero | thanks, ttl. | 21:37 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-os_nova master: Define condition for the first play host one time https://review.opendev.org/756332 | 21:44 |
| openstackgerrit | Merged openstack/openstack-ansible master: Update integrated ansible-lint rules https://review.opendev.org/756121 | 22:57 |
| *** watersj has quit IRC | 23:01 | |
| *** CeeMac has quit IRC | 23:01 | |
| *** tosky has quit IRC | 23:15 | |
| *** nurdie has quit IRC | 23:20 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_heat master: Trigger uwsgi restart https://review.opendev.org/756724 | 23:23 |
| *** nurdie has joined #openstack-ansible | 23:36 | |
| *** nurdie has quit IRC | 23:40 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!