*** macz_ has quit IRC | 00:27 | |
*** macz_ has joined #openstack-ansible | 00:52 | |
*** spatel has quit IRC | 00:53 | |
*** macz_ has quit IRC | 00:56 | |
*** macz_ has joined #openstack-ansible | 01:13 | |
*** macz_ has quit IRC | 01:18 | |
*** maharg101 has joined #openstack-ansible | 01:38 | |
*** maharg101 has quit IRC | 01:43 | |
*** macz_ has joined #openstack-ansible | 01:55 | |
*** macz_ has quit IRC | 01:59 | |
*** macz_ has joined #openstack-ansible | 03:02 | |
*** spatel has joined #openstack-ansible | 03:04 | |
*** macz_ has quit IRC | 03:07 | |
*** cshen has quit IRC | 03:20 | |
*** macz_ has joined #openstack-ansible | 03:23 | |
*** macz_ has quit IRC | 03:28 | |
*** macz_ has joined #openstack-ansible | 03:44 | |
*** macz_ has quit IRC | 03:49 | |
*** EZplay has joined #openstack-ansible | 04:09 | |
*** cshen has joined #openstack-ansible | 04:12 | |
*** cshen has quit IRC | 04:17 | |
*** EZplay has quit IRC | 04:19 | |
*** macz_ has joined #openstack-ansible | 04:26 | |
*** macz_ has quit IRC | 04:30 | |
*** csmart has quit IRC | 04:37 | |
*** lemko has quit IRC | 04:37 | |
*** gillesMo has quit IRC | 04:37 | |
*** csmart has joined #openstack-ansible | 04:42 | |
*** lemko has joined #openstack-ansible | 04:42 | |
*** gillesMo has joined #openstack-ansible | 04:42 | |
*** ioni has quit IRC | 04:45 | |
*** fridtjof[m] has quit IRC | 04:45 | |
*** masterpe has quit IRC | 04:45 | |
*** manti has quit IRC | 04:46 | |
*** macz_ has joined #openstack-ansible | 04:46 | |
*** csmart has quit IRC | 04:49 | |
*** macz_ has quit IRC | 04:51 | |
*** poopcat has quit IRC | 04:54 | |
*** poopcat has joined #openstack-ansible | 04:57 | |
*** openstackstatus has quit IRC | 04:58 | |
*** openstack has joined #openstack-ansible | 05:01 | |
*** ChanServ sets mode: +o openstack | 05:01 | |
*** fridtjof[m] has joined #openstack-ansible | 05:12 | |
*** ioni has joined #openstack-ansible | 05:15 | |
*** manti has joined #openstack-ansible | 05:18 | |
*** spatel has quit IRC | 05:19 | |
*** csmart has joined #openstack-ansible | 05:19 | |
*** evrardjp has quit IRC | 05:33 | |
*** evrardjp has joined #openstack-ansible | 05:33 | |
*** cshen has joined #openstack-ansible | 06:14 | |
*** cshen has quit IRC | 06:18 | |
*** gyee has quit IRC | 06:54 | |
*** cshen has joined #openstack-ansible | 07:03 | |
*** macz_ has joined #openstack-ansible | 07:27 | |
*** miloa has joined #openstack-ansible | 07:31 | |
*** macz_ has quit IRC | 07:32 | |
*** rpittau|afk is now known as rpittau | 07:51 | |
*** maharg101 has joined #openstack-ansible | 08:13 | |
*** andrewbonney has joined #openstack-ansible | 08:21 | |
jrosser | morning | 08:25 |
---|---|---|
noonedeadpunk | o/ | 08:25 |
jrosser | noonedeadpunk: i filed a bug for mariadb https://jira.mariadb.org/browse/MDEV-24829 | 08:25 |
jrosser | seems like we need a stack trace | 08:25 |
CeeMac | \o | 08:26 |
* noonedeadpunk wondering when 10.5.9 is going to be released | 08:27 | |
noonedeadpunk | I can hardly imagine catching this with trace tbh | 08:29 |
noonedeadpunk | or we need to always start with trace in CI | 08:29 |
noonedeadpunk | and hopefully catch it one day | 08:29 |
jrosser | trouble is it doesnt crash | 08:30 |
jrosser | so i think what they want is attach with gdb when it's locked up and then `thread apply all bt` | 08:31 |
jrosser | i was going to try a few focal AIO to see if i can catch it | 08:32 |
noonedeadpunk | let me spawn severl as well then to increase chances | 08:32 |
jrosser | theres a link in the last jira comment to how they expect traces to be made | 08:33 |
noonedeadpunk | I did it several times but already clean forgot... | 08:34 |
noonedeadpunk | and it was not very obvious even having this doc iirc | 08:34 |
noonedeadpunk | yeah, because on ubuntu binary is always stripped and there's no debug version... | 08:40 |
* noonedeadpunk regrets about descision to spawn vms | 08:42 | |
jrosser | argh | 08:44 |
jrosser | they have a tar here https://downloads.mariadb.org/mariadb/10.5.8/ | 08:46 |
jrosser | i wonder if it's enough to run galera playbook / remove package constantly till it breaks | 08:46 |
jrosser | noonedeadpunk: here are the debug pacakges http://mirrors.coreix.net/mariadb/repo/10.5/ubuntu/dists/focal/main/debug/binary-amd64/Packages | 08:49 |
jrosser | even though the binaries are stripped the symbols are shipped in seperate packages | 08:49 |
jrosser | gdb is smart enough to find them at runtime, theres some sort of ID built into the binary which allows it to know which symbols to load | 08:50 |
noonedeadpunk | Is it some extra repo? Just can't find these with apt search :( | 08:51 |
jrosser | usually it's needed to add an extra apt repo | 08:51 |
jrosser | sudo add-apt-repository 'deb [arch=amd64] http://mirrors.coreix.net/mariadb/repo/10.5/ubuntu focal main/debug | 09:00 |
noonedeadpunk | aha, yep, thanks. just needed to add another one with main/debug path | 09:07 |
noonedeadpunk | eventually it's even documented https://downloads.mariadb.org/mariadb/repositories/# | 09:09 |
noonedeadpunk | we should have made `galera_repo` a list of repos... | 09:11 |
*** tosky has joined #openstack-ansible | 09:12 | |
jrosser | i think i'm going to worry about adding the repo when/if i get a wedged up galera | 09:13 |
openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: DNM - test case for mariadb failure to start https://review.opendev.org/c/openstack/openstack-ansible/+/774782 | 09:17 |
*** macz_ has joined #openstack-ansible | 09:28 | |
*** macz_ has quit IRC | 09:33 | |
openstackgerrit | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_keystone master: DNM - Check reason for DB failure on focal https://review.opendev.org/c/openstack/openstack-ansible-os_keystone/+/774788 | 09:40 |
*** macz_ has joined #openstack-ansible | 09:49 | |
*** macz_ has quit IRC | 09:54 | |
*** macz_ has joined #openstack-ansible | 10:10 | |
*** macz_ has quit IRC | 10:15 | |
jrosser | noonedeadpunk: http://paste.openstack.org/show/802502/ | 10:30 |
jrosser | i got one :) | 10:30 |
noonedeadpunk | oh, sweet | 10:31 |
noonedeadpunk | I just finished setup-hosts and got distracted :( | 10:31 |
jrosser | i did about 20 rounds of apt-purge mariadb-common and then re-run the galera playbook in a metal AIO, no errors | 10:31 |
jrosser | then tried an LXC and second time i destroyed the container and re-created it did that | 10:32 |
jrosser | i wonder if uninstalling the package is sufficient though | 10:32 |
noonedeadpunk | I'm wondering if we're doing smth strange... Ie I see `mysql_parse (thd=thd@entry=0x7fc838000c58, rawbuf=rawbuf@entry=0x7fc8380106f0 "CREATE USER 'monitoring'@'localhost'", length=length@entry=36,` here, and I saw also https://zuul.opendev.org/t/openstack/build/b8f7aff4fb234a9689710ff0338d6a4e/log/logs/host/mariadb.service.journal-23-27-22.log.txt#213 | 10:36 |
noonedeadpunk | that log is from failed db creation on keystone | 10:36 |
jrosser | is that for the xinetd check script? | 10:37 |
noonedeadpunk | haven't checked yet, but might be, yes | 10:37 |
jrosser | could easily be that haproxy is checking the galera status at the same time that user is being made | 10:37 |
*** ioni has quit IRC | 10:42 | |
openstackgerrit | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add hosts integrated tests https://review.opendev.org/c/openstack/openstack-ansible/+/774685 | 10:45 |
noonedeadpunk | I'm wondering if we should start using maxscale one day instead of haproxy to lb mysql requests... | 10:46 |
noonedeadpunk | (in case this might be related) | 10:47 |
*** ioni has joined #openstack-ansible | 10:53 | |
*** ianychoi__ is now known as ianychoi | 11:30 | |
openstackgerrit | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_keystone master: DNM - Check reason for DB failure on focal https://review.opendev.org/c/openstack/openstack-ansible-os_keystone/+/774788 | 11:31 |
*** cloudnull has quit IRC | 12:02 | |
*** cloudnull has joined #openstack-ansible | 12:05 | |
jrosser | noonedeadpunk: we could use something like this to inhibit the cluster check whilst restarting mariadb https://github.com/openstack/openstack-ansible-galera_server/blob/master/templates/clustercheck.j2#L20-L32 | 12:32 |
*** mgariepy has quit IRC | 12:46 | |
*** cshen has quit IRC | 12:57 | |
*** cshen has joined #openstack-ansible | 12:59 | |
*** macz_ has joined #openstack-ansible | 13:07 | |
openstackgerrit | Merged openstack/ansible-role-python_venv_build master: Import wheels build only when necessary https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/774159 | 13:11 |
*** macz_ has quit IRC | 13:11 | |
openstackgerrit | Merged openstack/ansible-role-python_venv_build master: Remove preflight checks https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/774420 | 13:13 |
*** mgariepy has joined #openstack-ansible | 13:14 | |
*** macz_ has joined #openstack-ansible | 13:27 | |
*** macz_ has quit IRC | 13:32 | |
*** jpvlsmv has joined #openstack-ansible | 13:42 | |
*** jpvlsmv has quit IRC | 13:46 | |
*** jpvlsmv has joined #openstack-ansible | 13:47 | |
*** macz_ has joined #openstack-ansible | 13:48 | |
*** macz_ has quit IRC | 13:53 | |
*** cloudnull has quit IRC | 13:56 | |
*** spatel has joined #openstack-ansible | 13:57 | |
*** cloudnull has joined #openstack-ansible | 13:59 | |
openstackgerrit | Dmitriy Rabotyagov proposed openstack/ansible-role-python_venv_build stable/victoria: Import wheels build only when necessary https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/774804 | 14:03 |
admin0 | anyone seen this error before: Error in store configuration. Adding images to store is disabled.: glance_store.exceptions.StoreAddDisabled: Configuration for store failed. .. | 14:05 |
admin0 | its glance over NAS | 14:06 |
admin0 | same config working fine in another cluster :( | 14:06 |
admin0 | i validated that it has mounted the nfs path and i can write to it as well | 14:06 |
*** ianychoi has quit IRC | 14:09 | |
*** macz_ has joined #openstack-ansible | 14:09 | |
*** jpvlsmv has quit IRC | 14:10 | |
*** macz_ has quit IRC | 14:14 | |
spatel | admin0 what is default_store= ? | 14:14 |
admin0 | https://gist.githubusercontent.com/a1git/d1a03bbb3bd97b0063927e95a84ca41a/raw/ab0e1b0eb41e2086efdef73a6bf9cc5a9fa12a1f/gistfile1.txt | 14:15 |
admin0 | let me fetch that | 14:15 |
openstackgerrit | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add hosts integrated tests https://review.opendev.org/c/openstack/openstack-ansible/+/774685 | 14:16 |
admin0 | spatel, there is no default_store anywhere in the config .. this is 22.0.0 .. i have another 21.1.0 where i am using nfs for glance and I could not find this variable there as well | 14:17 |
spatel | admin0 also make sure /var/lib/glance/images own by glance | 14:18 |
admin0 | enabled_backends = file:file,http:http,cinder:cinder | 14:18 |
admin0 | that is there | 14:18 |
admin0 | its nfs and its at 777 | 14:19 |
admin0 | so glance can read-write | 14:19 |
jrosser | spatel: do you have any centos networking config examples for this? http://lists.openstack.org/pipermail/openstack-discuss/2021-February/020390.html | 14:25 |
spatel | Yes i saw that in mailing list and going to reply :) | 14:25 |
jrosser | even better we can make a patch | 14:26 |
spatel | i thought we are going to tell folks move on from CentOS 8 | 14:26 |
jrosser | well yes indeed | 14:26 |
spatel | Its end of life and small fix won't last for long time. | 14:26 |
spatel | let me reply anyway and see | 14:28 |
jrosser | i also get the impression that the ML post wants a "kitchen sink" installer that does everything | 14:28 |
admin0 | could it be due to this ? | Unable to create trust: no such option collect_timing in group [keystone_authtoken] Use the existing user token. -- .. not sure .. all commands to list etc work fine ( i see no errors ) | 14:34 |
admin0 | playbook runs successfully without isues | 14:34 |
spatel | admin0 I had similar issue yesterday when i was setting up ceph storage for glance and my issue was glance was throwing permission error | 14:43 |
admin0 | i found somewhere that i cannot mount the same folder in 2 diferent glance | 14:44 |
admin0 | so checking that | 14:44 |
spatel | If you are setting this up for NFS then all i can see glance not able to access NFS storage or permission issue. NFS is tricky sometime | 14:44 |
spatel | NFS has lots of locking issue if you don't use NFSv4 | 14:44 |
*** macz_ has joined #openstack-ansible | 14:47 | |
spatel | jrosser don't you think we should put warning here or redhat official link for centos 8 deployment (that this is EOL soon and support is experimental) | 14:47 |
spatel | https://docs.openstack.org/openstack-ansible/latest/user/aio/quickstart.html | 14:47 |
*** macz_ has quit IRC | 14:51 | |
*** macz_ has joined #openstack-ansible | 15:08 | |
*** macz_ has quit IRC | 15:13 | |
*** mgariepy has quit IRC | 15:15 | |
*** macz_ has joined #openstack-ansible | 15:29 | |
openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: Add documentation warning about Centos-8 end-of-life https://review.opendev.org/c/openstack/openstack-ansible/+/774887 | 15:31 |
jrosser | spatel: ^ | 15:31 |
spatel | +1 | 15:32 |
*** macz_ has quit IRC | 15:33 | |
jrosser | i need to have another go on centos-stream | 15:33 |
spatel | Do we have CI job for stream? | 15:33 |
jrosser | but the only route i can see is upgrading the entire of systemd from EPEL | 15:34 |
spatel | I can build LAB for centOS-stream if you want to test it out outside of CI | 15:34 |
jrosser | and i'm not really sure if taking such a critical component from there is wise | 15:34 |
spatel | hmm | 15:34 |
jrosser | all for one package that RH refuse to support, it's really sad :( | 15:35 |
spatel | RH trying to make everything highly commercialized | 15:36 |
*** mgariepy has joined #openstack-ansible | 15:44 | |
*** macz_ has joined #openstack-ansible | 15:49 | |
admin0 | ok .. issue solved .. new stuff learnt .. you cannot mount the same glance folder via nfs in 2 different openstack clusters ( trying to migrate from old to new) | 15:51 |
*** macz_ has quit IRC | 15:53 | |
*** djhankb has quit IRC | 15:55 | |
*** djhankb has joined #openstack-ansible | 15:56 | |
spatel | admin0 that is not good | 16:02 |
spatel | may be you need this - https://www.stackalytics.com/report/blueprint/glance/glance-multifilesystem-store | 16:04 |
admin0 | i was trying to migrate like 1 TB of images by adding the same store and just rename the uuid from old -> new .. but now i have to copy it to a new folder | 16:05 |
admin0 | still ok | 16:05 |
admin0 | that error was cryptic though | 16:05 |
*** fanfi has joined #openstack-ansible | 16:08 | |
*** macz_ has joined #openstack-ansible | 16:10 | |
fanfi | hi, could someone help me please? After fresh OSA deployment i do not have any DHCP agent. Which parameter is needed to setup it ? | 16:13 |
spatel | fanfi did you use AIO deployment? | 16:18 |
fanfi | no | 16:18 |
fanfi | 3node setup +compute node | 16:19 |
fanfi | with Open vSwitch agent | 16:19 |
spatel | DHCP agent run on infra nodes on metal | 16:19 |
*** LowKey has joined #openstack-ansible | 16:20 | |
spatel | you can see them somewhere on this path /openstack/venvs/neutron-21.0.1.dev4/etc/neutron/ | 16:20 |
fanfi | it is not a part of neutron container ? | 16:20 |
spatel | no | 16:20 |
openstackgerrit | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add hosts integrated tests https://review.opendev.org/c/openstack/openstack-ansible/+/774685 | 16:21 |
spatel | DHCP agent get install on Physical host on infra node | 16:21 |
spatel | neutron container only run neutron-server service | 16:21 |
fanfi | do i need to specify someting for dhcp agent in variables ? | 16:23 |
spatel | no | 16:23 |
spatel | it should be there | 16:23 |
fanfi | i can't find any config for neutron on infra nodes | 16:26 |
fanfi | i am using ubuntu 20 | 16:26 |
fanfi | and source install method | 16:27 |
spatel | look at here - http://paste.openstack.org/show/802527/ | 16:27 |
spatel | OSA install components in /openstack/venv | 16:28 |
fanfi | in /openstack i have only log directory | 16:28 |
spatel | then something is wrong.. | 16:29 |
fanfi | http://paste.openstack.org/show/802528/ | 16:30 |
spatel | v-node01_neutron_ovn_northd_container | 16:31 |
spatel | did you use OVN to deploy? | 16:31 |
spatel | jrosser may be you can help here. | 16:32 |
fanfi | no i am use "openvswitch" | 16:32 |
spatel | OVN use OpenvSwitch | 16:32 |
spatel | I don't have experience with OVN so hope someone else chime in | 16:35 |
fanfi | oki, thank you | 16:35 |
noonedeadpunk | id say it's more mgariepy who played with ovn | 16:36 |
mgariepy | i played briefly with it but i plan to work on it when i have time, | 16:44 |
fanfi | which neutron plugin you are using for ? | 16:46 |
mgariepy | currently my setup is runing with ovs, i do plan to switch to ovn at some point (once i have time to fix a few things) | 16:51 |
mgariepy | i don't use much plugins only the default one for now. | 16:53 |
mgariepy | but with the openvswitch firewall | 16:53 |
spatel | fanfi why did you decided to use OVN (just curious) | 17:03 |
mgariepy | http://paste.openstack.org/show/802531/ | 17:03 |
mgariepy | shiny and new. | 17:04 |
mgariepy | mostly to get rid of the annoyong agent process, | 17:04 |
mgariepy | i do have like 400-500 project on my cloud and the falback when a node dies is annoyingly long. | 17:05 |
mgariepy | ovn seems to be better on paper | 17:05 |
mgariepy | also, i work in a university for research purposes. | 17:06 |
mgariepy | ovs config firewall_driver: openvswitch | 17:07 |
*** LowKey has quit IRC | 17:07 | |
openstackgerrit | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add hosts integrated tests https://review.opendev.org/c/openstack/openstack-ansible/+/774685 | 17:14 |
openstackgerrit | Dmitriy Rabotyagov proposed openstack/openstack-ansible-openstack_hosts master: Use integrated tests https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/774688 | 17:16 |
fanfi | i would like to use OVS, not ovn | 17:26 |
fanfi | that my setup http://paste.openstack.org/show/802534/ | 17:29 |
*** LowKeys has joined #openstack-ansible | 17:34 | |
*** d34dh0r53 has quit IRC | 17:43 | |
*** d34dh0r53 has joined #openstack-ansible | 17:45 | |
mgariepy | fanfi, what is your issue ? | 17:45 |
*** maharg101 has quit IRC | 17:46 | |
mgariepy | fanfi, on your infra host , do you have openvswith running ? | 17:49 |
mgariepy | fanfi, ovs-vsctl show does it show you something that seems to be ok / | 17:50 |
mgariepy | ? | 17:50 |
*** jbadiapa has quit IRC | 17:51 | |
spatel | mgariepy he was asking about where is DHCP agent because he didn't find anything related neutron on infra* node | 17:53 |
spatel | when i noticed OVN container in his paste so got confused | 17:54 |
LowKeys | Hi, morning.. how to fix this type of error ? http://paste.openstack.org/show/QmV5uL23pLeYboeyD6Wt/ | 17:55 |
jrosser | LowKeys: you'll need to paste more of the log than that becasue the file it cannot find should have been created previously | 18:00 |
openstackgerrit | wes hayutin proposed openstack/openstack-ansible-os_tempest master: remove master standalone upgrade https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/774916 | 18:01 |
mgariepy | ovn container maybe you have a mis-configured stuff. | 18:01 |
openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-galera_server master: Move installation of xinetd clustercheck until after mariadb is installed and configured https://review.opendev.org/c/openstack/openstack-ansible-galera_server/+/774918 | 18:11 |
LowKeys | jrosser: i re-running playbooks with "-e venv_rebuild=yes" and it's solved the issues.. | 18:12 |
jrosser | noonedeadpunk: ^ galera role does really strange things with handlers, it looks like clustercheck is installed way before the db is boostrapped and the users created, hopefully my change makes the makes the clustercheck stuff be done at the end | 18:13 |
jrosser | LowKeys: as you are running 20.0.0 you should know about https://bugs.launchpad.net/bugs/1914301 | 18:15 |
openstack | Launchpad bug 1914301 in openstack-ansible "passing venv_rebuild=true leaves repo server in unusable state" [Undecided,Fix released] | 18:15 |
*** miloa has quit IRC | 18:16 | |
LowKeys | jrosser: ok noted, my bad.. | 18:20 |
*** rpittau is now known as rpittau|afk | 18:23 | |
*** ioni has quit IRC | 18:38 | |
*** spatel has quit IRC | 18:40 | |
*** spatel has joined #openstack-ansible | 18:44 | |
*** cshen has quit IRC | 19:35 | |
*** andrewbonney has quit IRC | 19:39 | |
*** maharg101 has joined #openstack-ansible | 19:43 | |
fanfi | mgariepy, hi as spatel sed. I do not see any dhcp agent running on my new infra | 19:45 |
spatel | fanfi i would say you should remove OVN piece if you don't know about that | 19:46 |
*** maharg101 has quit IRC | 19:47 | |
fanfi | okay, but I have no idea where I put OVN part in my config | 19:49 |
mgariepy | did you paste your config somewhere? | 19:49 |
mgariepy | fanfi, can we check this tomorrow ? | 19:53 |
spatel | he might copy that from example (may got confused between OVN vs OVS) | 19:53 |
fanfi | http://paste.openstack.org/show/802545/ | 19:54 |
mgariepy | i'm usualy there from 9h-ish to 15h-ish EST. | 19:54 |
*** jpvlsmv has joined #openstack-ansible | 19:58 | |
mgariepy | fanfi, also i would need your /etc/openstack_deploy/env.d/* | 20:01 |
*** maharg101 has joined #openstack-ansible | 20:03 | |
fanfi | oh...i see there are config for OVN | 20:04 |
fanfi | i put it to the paste | 20:05 |
fanfi | it could be just deleted ? | 20:05 |
mgariepy | the inventory also need to be updated. | 20:06 |
mgariepy | adding is pretty much seamless. substraction not so much. | 20:06 |
spatel | if it going to be production environment and you just started building then i would say destroy everything and rebuild again.. (i would do that if its my cloud) | 20:07 |
fanfi | och...okay thank you ...i will try it again :) | 20:08 |
spatel | you never know what goes where in config.. until you are expert | 20:08 |
fanfi | an i have one more question, how I can eddit this parameter for LXC host --dhcp-option="6,${LXC_ADDR}" that create new def route with i do not want to use. for my lxc containers | 20:14 |
*** ioni has joined #openstack-ansible | 20:17 | |
fanfi | in file /usr/local/bin/lxc-system-manage | 20:17 |
*** gyee has joined #openstack-ansible | 20:21 | |
openstackgerrit | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add hosts integrated tests https://review.opendev.org/c/openstack/openstack-ansible/+/774685 | 20:23 |
openstackgerrit | Dmitriy Rabotyagov proposed openstack/openstack-ansible-galera_server master: Replace upgrade jobs with infra_lxc https://review.opendev.org/c/openstack/openstack-ansible-galera_server/+/775011 | 20:24 |
*** maharg101 has quit IRC | 20:37 | |
*** jpvlsmv has quit IRC | 21:15 | |
*** jpvlsmv has joined #openstack-ansible | 21:29 | |
openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-galera_server master: Move installation of xinetd clustercheck until after mariadb is installed and configured https://review.opendev.org/c/openstack/openstack-ansible-galera_server/+/774918 | 21:31 |
openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-galera_server master: Install xinetd clustercheck until after mariadb is installed and setup https://review.opendev.org/c/openstack/openstack-ansible-galera_server/+/774918 | 21:33 |
openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-galera_server master: Install xinetd clustercheck after mariadb is installed and setup https://review.opendev.org/c/openstack/openstack-ansible-galera_server/+/774918 | 21:33 |
*** jpvlsmv has quit IRC | 21:34 | |
*** LowKeys has quit IRC | 22:03 | |
*** spatel has quit IRC | 22:09 | |
*** jpvlsmv has joined #openstack-ansible | 22:38 | |
*** eatthoselemons has joined #openstack-ansible | 22:39 | |
eatthoselemons | I am trying to do my first openstack ansible install and am trying to figure out the best storage configuration for my situation, I was wondering if this would be the correct channel to ask in or if I should try one of the more specific channels like "openstack-swift"? | 22:48 |
jrosser | here is a good place to ask | 22:52 |
jrosser | generally people doing deployments are here, rather than purely developing the services | 22:52 |
eatthoselemons | okay good to know | 22:56 |
eatthoselemons | So I have a single storage node (all compute nodes don't have their own storage) | 22:57 |
eatthoselemons | with that I would like to put, glance, swift, cinder and manila on there | 22:57 |
eatthoselemons | (which I believe I need for an openstack config | 22:57 |
eatthoselemons | ) | 22:57 |
eatthoselemons | well need object storage for my application, and glance since there is no storage on compute nodes | 22:58 |
jrosser | swift and cinder are optional | 22:58 |
jrosser | you'd need glance regardless of storage on the compute nodes | 22:58 |
jrosser | *swift and manila sorry | 22:58 |
eatthoselemons | Okay so sounds like I have to have, glance, swift, and cinder (since swift is optional but need it for my application) | 22:59 |
eatthoselemons | Is it possible to install all of those to the same physical machine without vm's? | 23:00 |
jrosser | you have many many choices | 23:01 |
eatthoselemons | (with the note of my understanding is for openstack-ansible you need to have the storage already setup. ie openstack-ansible doesn't setup the storage for you | 23:01 |
jrosser | LVM backing for cinder works pretty much out of the box | 23:01 |
jrosser | you can use NFS for glance which would match nicely with a single host | 23:02 |
eatthoselemons | for nfs you are saying create a volume and expose it via nfs and then on the same machine run glance? | 23:03 |
jrosser | glance is an API that runs on the controller | 23:03 |
jrosser | there are choices for what the backing store is | 23:04 |
jrosser | you'd have to consult the glance documentation | 23:04 |
jrosser | similar for cinder, cinder is an API on the controller | 23:05 |
jrosser | cinder-volume is a service which can run on a controller (or elsewhere) and integrate various storage backends | 23:05 |
eatthoselemons | ah I didn't know glance was an api on the controller good to know | 23:06 |
jrosser | imho you should start with an openstack-ansible all-in-one and understand some of the architecture | 23:07 |
eatthoselemons | so when you are inputting the ip's for cinder/glance then you are not putting the storage node with the storage but where to put the api's? | 23:07 |
jrosser | in general terms, yes | 23:08 |
eatthoselemons | My understanding was that the all-in-one would just set everything up for you so you wouldn't understand the architecture, ie would be good for learning to do things on openstack but not for learning openstack itself | 23:08 |
eatthoselemons | I take it that understanding is incorrect? | 23:08 |
jrosser | if you want to understand how openstack-ansible works then you should start with the all-in-one | 23:08 |
jrosser | and have one available at all times as you try to build a proper deployment to use as a reference | 23:09 |
eatthoselemons | openstack ansible is not its own thing though right? It is just a set of install scripts for openstack? | 23:09 |
eatthoselemons | ie the goal would be to understand openstack through openstack-ansible? | 23:09 |
jrosser | i assume you're here asking about making a deployment with openstack-ansible? | 23:10 |
jrosser | https://github.com/openstack/openstack-ansible | 23:10 |
eatthoselemons | I would like an openstack deployment, and from my understanding the best way to get that is to use openstack-ansible to deploy openstack | 23:10 |
eatthoselemons | best way being the easier/don't have to be an expert in all things openstack | 23:11 |
*** poopcat has quit IRC | 23:11 | |
*** poopcat has joined #openstack-ansible | 23:13 | |
jrosser | openstack-ansible is more than "just a set of install scripts", becasue you end up with a fairly well battle tested architecture proven in some large deployments | 23:14 |
eatthoselemons | they are sane defaults correct? | 23:15 |
jrosser | thats the idea | 23:15 |
eatthoselemons | so with openstack-ansible you are less likely to configure something incorrectly? | 23:15 |
jrosser | it's kind of the opposite of a shrink-wrap software installer | 23:16 |
eatthoselemons | How is it the opposite? | 23:16 |
jrosser | everything has sensible defaults, and a reference deployment architecture that you can choose to follow if you wish | 23:16 |
jrosser | on the other hand absolutely everything is configurable so it's totally possible to make something broken | 23:17 |
jpvlsmv | I'll attest that it's easy to make something broken :) | 23:17 |
eatthoselemons | ah you are saying that it is the opposite of a software installer in that everything is configurable? | 23:18 |
eatthoselemons | vim | 23:18 |
eatthoselemons | ehh wrong window | 23:18 |
jrosser | if you want some sort of wizard type setup tool then it's not that | 23:19 |
jpvlsmv | well. aio is pretty close to a wizard-type | 23:19 |
jrosser | true, which is why i suggest that is a great starting point | 23:19 |
eatthoselemons | that makes sense | 23:20 |
jrosser | fwiw the all-in-one is the exact code we run dozens of times a day in CI to validate every code change | 23:20 |
jrosser | so it's very well tested | 23:20 |
eatthoselemons | I did see that the aio is the method for testing | 23:21 |
eatthoselemons | is there something I should do with the aio install to learn? or just kinda poke around? | 23:21 |
jrosser | you've asked quite a bit about storage, so maybe understand how the AIO does that and see how it would map to what you want to do | 23:22 |
jrosser | theres obviously no H/A in the AIO, so deciding your approach to that (if you want high-availability control plane) is also a good excercise | 23:23 |
*** kleini has quit IRC | 23:25 | |
eatthoselemons | for that I assume I should also undertand the bootstrap-aio.sh script? | 23:26 |
eatthoselemons | I don't need any H/A so don't have to do too much configuration | 23:26 |
jrosser | bootstrap-aio.sh is of limited use beyond the AIO | 23:26 |
jrosser | stepping beyond the AIO to your own setup the idea is you do the preparation of the target hosts in whichever way you prefer | 23:27 |
eatthoselemons | ah, is there some code that I can go through to see what the aio uses for its config? or is it all spread out in defines throughout the repo? | 23:27 |
jrosser | generally places have their own pxeboot/foreman/maas/whatever | 23:27 |
jrosser | there is an ansible role in tests/roles/boostrap-host which brings preps everything for the AIO | 23:28 |
eatthoselemons | The machines I will have all are static (just 1 compute for example) so don't expect to need maas or pxeboot, is that a reasonable line of thinking? | 23:28 |
jrosser | thats fine | 23:29 |
eatthoselemons | ah okay I will start going through those anisble roles | 23:29 |
jrosser | https://docs.openstack.org/project-deploy-guide/openstack-ansible/victoria/targethosts.html | 23:30 |
eatthoselemons | sounds like a plan start with the aio, and poke around understand the bootstrap-host role | 23:31 |
eatthoselemons | are there anything I should know that are weird/poorly explained? | 23:31 |
jrosser | maybe don't underestimate the amount if learning to do | 23:32 |
eatthoselemons | ah I just totally skipped the configuring the storage step, I was thinking of using zfs so just skipped the lvm portion, I see how it applys now | 23:33 |
eatthoselemons | Yea I tend to rush into things, and underestimate them | 23:33 |
eatthoselemons | for the aio it has the horizon dashboard correct? | 23:34 |
jrosser | if you take the default option of deploying with LXC containers, yes | 23:36 |
eatthoselemons | okay good to know | 23:36 |
eatthoselemons | and the openstack ansible deploys to bare metal though right? (well the controller services are in containers if I recall correctly, been a bit since I looked at the controller node stuff) | 23:37 |
jrosser | for the compute nodes everything is just on the metal | 23:41 |
eatthoselemons | ah okay | 23:42 |
eatthoselemons | So another question that I don't think the aio will cover, (or maybe it does) since vm's wont have local storage then the speed of the glance service is very important, it looks like I could use infiniband to be the boot drives for the vm's | 23:45 |
eatthoselemons | is there anything in the AIO that I should pay attention to for that? | 23:45 |
eatthoselemons | or is that all just done on the openstack ansible config? | 23:45 |
eatthoselemons | and won't learn anything about that from the aio? | 23:46 |
jrosser | glance just stores images | 23:51 |
*** CeeMac has quit IRC | 23:51 | |
eatthoselemons | does that mean that all userfiles are also loaded into memory? I assumed you could have everything not the kernel be stored on the storage to save on memory footprint of the vm? | 23:55 |
eatthoselemons | based on the "glance just stores images" I take it that that understanding is incorrect? | 23:55 |
eatthoselemons | Another way to put that is I keep seeing that fast storage is useful for vm's I assumed that was for packages/userfiles/etc | 23:56 |
jrosser | glance is an image store and it's where you put your ubuntu/centos/whatever cloud images | 23:58 |
jrosser | when you boot them they're transferred to whatever storage is backing the VM | 23:58 |
jrosser | which may be local to the compute node or elsewhere | 23:58 |
eatthoselemons | if the compute node doesn't have any storage then where are they transfered to? | 23:59 |
jrosser | and can be many protocols, iscsi, NFS, ceph rbd | 23:59 |
jrosser | if you use cinder for block storage, whichever storage you choose to integrate with cinder-volume | 23:59 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!