*** KeithMnemonic has quit IRC | 00:30 | |
*** mbindlish has joined #openstack-monasca | 03:34 | |
*** viks____ has joined #openstack-monasca | 04:26 | |
*** nikparasyr has joined #openstack-monasca | 06:44 | |
*** dougsz has joined #openstack-monasca | 07:16 | |
*** witek_ has joined #openstack-monasca | 08:14 | |
*** k_mouza has joined #openstack-monasca | 08:39 | |
*** k_mouza has quit IRC | 09:47 | |
*** k_mouza has joined #openstack-monasca | 09:58 | |
nikparasyr | good morning :). continuing the monasca HA setup journey. I have managed to have each persister pick-up all the messages from kafka. defining a different group_id was not enough, I also had to set: legacy_kafka_client_enabled = false. With these settings it works as I would expect. Now i'm wondering if the metrics topic is the only one I I should have different group_ids. I'm checking this post: https://www.stackhpc.com/monasc | 10:22 |
---|---|---|
nikparasyr | a-comes-to-kolla.html (btw rly nice post dougsz) to better understand the monasca layout. From that graph that consumers on log related topics dont need different group_ids, same for notification related topics. There are still 2 topics I'm not sure about. These are the events and alarm-state-transitions topics. In my understanding the event topic functionality is still under developement. For the alarm-state-transitions i see | 10:22 |
nikparasyr | here: https://review.opendev.org/#/c/738585/2 that you also add a different group_id. Can you elaborate a bit on the functionality behind the topic etc? | 10:22 |
nikparasyr | dont want to end up having 3 alarms for the same incident as 3 transitions happened etc | 10:22 |
dougsz | good morning! | 10:23 |
dougsz | interesting - so some issue with the old python kafka client, legacy_kafka_client_enabled = false should probably be the default now | 10:24 |
nikparasyr | this is also a setting that i havent found on all monasca services. monasca-persister, notification, thresh and api have this setting, the others not | 10:26 |
dougsz | group_id for monasca_alarm_state_transitions_topic is just for the alarm history which can be stored in InfluxDB | 10:32 |
dougsz | glad the blog post was helpful :) | 10:32 |
nikparasyr | if that is the case it should be replicated as well. Thanks :) | 10:38 |
dougsz | Good question about the kafka client setting in other services - need to remind myself of what happened with the new client there | 10:40 |
witek_ | persister, api and notification support the new client | 10:42 |
witek_ | and yes, we should change the default value | 10:42 |
dougsz | thanks witek, does that need to be enabled in the config for other services, or is it default? | 10:42 |
witek_ | all services use old client per default | 10:43 |
dougsz | nikparasyr: yeah, it's only the python based Monasca services that should have the setting. The exception is the log API which has been deprecated. | 11:05 |
nikparasyr | ok. Also right now I can see some consumers and some topics via zookeeper and others via bootstrap-server (in terms of using the kafka cli). I guess this is related to using the legacy or not client. Anywayz. moving forward :) thanks | 11:10 |
dougsz | In KA Ussuri we switched to the unified Monasca API. We need a follow up change in KA Ussuri onwards to disable the legacy client in other services, I think it works at moment because the Kafka messaging version is still pinned: https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/kafka/templates/kafka.server.properties.j2#L22 | 11:11 |
dougsz | Which release are you using again? | 11:12 |
hrw | morning | 11:14 |
hrw | 19:24 < hrw> https://tarballs.opendev.org/openstack/monasca-common/ lacks monasca-common-stable-ussuri.tar.gz tarball | 11:14 |
hrw | 19:24 < hrw> was this package abandoned after train? | 11:14 |
dougsz | morning hrw! | 11:15 |
dougsz | monasca-common - we are still using that | 11:15 |
hrw | but you do not make stable branch tarballs on CI. only master one | 11:16 |
dougsz | Maybe we are missing a release, I will check after lunch | 11:16 |
hrw | Kolla wants to move to stable branches for ussuri stable and monasca-common is one of those missing ones | 11:16 |
hrw | dougsz: release was done. I ask about monasca-common-stable-ussuri.tar.gz | 11:16 |
chaconpiza | I will take a look what we are missing the monasca-common-stable-ussuri.tar.gz tarball, thanks hrw for notice that | 11:17 |
hrw | chaconpiza: you miss some CI job iirc | 11:17 |
nikparasyr | dougsz: Im using train for monasca, openstack cluster is stein. monasca is HA standalone with external keystone | 11:41 |
chaconpiza | hrw, I think the issue could be on this line from monasca-common branch stable/ussuri: https://opendev.org/openstack/monasca-common/src/branch/stable/ussuri/.zuul.yaml#L6 | 11:48 |
chaconpiza | hrw: for example in monasca-notification we got this change from pengyuesheng https://opendev.org/openstack/monasca-notification/commit/88dbdcfb9e5effe0c31075ffe12533a3d28bbff3 | 11:53 |
hrw | chaconpiza: possible | 11:55 |
chaconpiza | :) | 11:56 |
hrw | which reminds me to check for victoria version of those jobs ;D | 11:57 |
hrw | project.yaml: - openstack-python3-victoria-jobs | 11:57 |
hrw | ok, kolla already migrated | 11:57 |
chaconpiza | I will send a change for monasca-common similar to pengyuesheng https://opendev.org/openstack/monasca-notification/commit/88dbdcfb9e5effe0c31075ffe12533a3d28bbff3 | 11:57 |
hrw | coolio | 11:58 |
chaconpiza | no to fajne. dzięki. | 11:59 |
*** zijlboot has quit IRC | 13:54 | |
*** zijlboot_ has joined #openstack-monasca | 13:54 | |
*** mbindlish has quit IRC | 15:13 | |
*** witek_ has quit IRC | 15:44 | |
*** k_mouza has quit IRC | 16:02 | |
*** nikparasyr has left #openstack-monasca | 16:20 | |
*** dougsz has quit IRC | 16:32 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!