#acumos-meeting: Acumos OA&M Project
Meeting started by bryan_att at 15:08:25 UTC
(full logs).
Meeting summary
-
- attendees Bryan, Krish, Nat, Parichay
(aimeeu,
15:09:50)
- conflicts, so Bryan will send out a doodle poll
to determine optimal meeting time (aimeeu,
15:11:14)
- #link
https://wiki.acumos.org/display/OAM/Meetings (aimeeu,
15:12:20)
- Krish: do we need any other repos? or will the
existing 2 repos be sufficient? (aimeeu,
15:12:52)
- Bryan: system-integration will be for platform
deployment, platform-oam will be for associated tools such as
logging (aimeeu,
15:14:02)
- Krish: should reports and/or a reporting tool
be part of platform-oam (aimeeu,
15:15:31)
- Parichay: backup and disaster recovery should
be part of platform-oam (aimeeu,
15:15:49)
- Bryan: good scope; we will evolve as more items
are incoporated (aimeeu,
15:17:02)
- Bryan shows list of not-closed Jira items for
system-integration and platform-oam (aimeeu,
15:17:54)
- Krish: let's focus on prioritizing - let's
focus on the High items (aimeeu,
15:20:33)
- Vishnu (vrov) joined the meeting (aimeeu,
15:21:11)
- #link
https://jira.acumos.org/browse/ACUMOS-1043 (aimeeu,
15:21:59)
- this encompasses back end; different companies
may wish to use different DBs; we should provide different options,
not exclude, and let operator decide what back-end components to
use (aimeeu,
15:23:35)
- Bryan: AIO script has ability to specify
pre-existing components, such as database, so database wouldn't be
installed (aimeeu,
15:24:27)
- AIO script lacks upgrade/migration (aimeeu,
15:24:46)
- Krish: Parichay - what type of storage do we
support, how do we specify different storage types - we need to work
through and create a good solution (aimeeu,
15:25:40)
- Bryan: k8s - important to have data live
outside the Docker containers (aimeeu,
15:26:16)
- persistent volumes (aimeeu,
15:26:31)
- Krish: some platforms, like Azure, provide load
balancing; Acumos needs to be more generic in how it uses a load
balancer (aimeeu,
15:27:10)
- Bryan: need to provide clear guidance; not all
load balancers support Kong - need to solve this reverse proxy with
different load balancers (aimeeu,
15:28:27)
- Krish: Parichay has been working on kong with
k8s (aimeeu,
15:29:18)
- Bryan: problem with kong in k8s cluster is that
you can't expose 8080 or 443 (aimeeu,
15:29:51)
- Bryan explains (aimeeu,
15:29:57)
- Krish - we've been working on this - need to
sync up (aimeeu,
15:30:24)
- Krish provides details (aimeeu,
15:30:43)
- discussion between Bryan and Krish; implication
is that the installation must have a load-balancer (aimeeu,
15:31:54)
- Krish: goal is to provide complete k8s with
nothing running outside (on docker ce for example) (aimeeu,
15:33:27)
- Kirsh: for k8s, should we pick a single
environment (Azure, bare metal, etc) and then expand to other
environments? (aimeeu,
15:34:42)
- Krish: for k8s, should we pick a single
environment (Azure, bare metal, etc) and then expand to other
environments? (aimeeu,
15:35:03)
- Bryan: k8s deployment of the platform should
not differ based on environment (aimeeu,
15:35:44)
- standing up a load-balancer may be environment
dependent but that's out of scope (aimeeu,
15:36:14)
- Krish points out that an environment such as
Azure provides a lot already set up compared to bare metal
(aimeeu,
15:37:01)
- Bryan: we should be pure cloud native - doesn't
matter where it's deployed, process should be the same (aimeeu,
15:37:52)
- Bryan: we will work collaboratively to address
these issues (aimeeu,
15:38:09)
- https://jira.acumos.org/browse/ACUMOS-1091 (aimeeu,
15:38:34)
- #link
https://jira.acumos.org/browse/ACUMOS-1091 (aimeeu,
15:38:51)
- Krish: we need get some of the ONAP ELK people
to take an active role in Acumos (aimeeu,
15:39:29)
- Ken: is this Sponden's area? (aimeeu,
15:39:45)
- Ken Kristiansen joined the meeting (aimeeu,
15:40:02)
- Krish: currently under platform-oam but if it
grows, it could be split into a separate project (aimeeu,
15:40:54)
- consensus that Sponden should be
involved (aimeeu,
15:41:18)
- Parichay has done a lot of work in this ELK
area (aimeeu,
15:42:02)
- Ken: Sponden has been leading the effort
between ONAP and Acumos ELK effort (aimeeu,
15:42:34)
- Ken: recommends Bryan have a call with
Sponden (aimeeu,
15:43:17)
- Bryan will reach out to Sponden to understand
what he's done and how he can help drive ELK in Acumos (aimeeu,
15:43:51)
- Krish: missing from Jira is functionality to
query database to build reports; we need to discuss this
further (aimeeu,
15:46:10)
- Krish: metrics examples: how many people logged
in, how many models uploaded, etc - reports that business managers
want (aimeeu,
15:46:52)
- Krish: need to be clear that business metrics
are needed and that they aren't based on physical resources
(aimeeu,
15:47:40)
- need to enhance logging capabilities to capture
what is currently missing, such as IP addresses of users
(aimeeu,
15:48:49)
- access logs of client IP addresses (aimeeu,
15:49:04)
- Ken: Parichay is already working on these
initiatives and is almost done (aimeeu,
15:50:21)
- Ken: pulling data directly from database is
only temporary until a solution is found (aimeeu,
15:51:22)
- Krish: need to set log retention (aimeeu,
15:51:40)
- need to give users choice of retention
period (aimeeu,
15:51:55)
- Bryan asks for help prioritizing Jira
tickets (aimeeu,
15:53:22)
- Bryan asks for who will focus on what
(aimeeu,
15:54:17)
- Logging: Parichay (aimeeu,
15:54:27)
- Bryan/Krish: Parichay should be made a
committer in system-integration (aimeeu,
15:56:22)
- Question about Amit Mishra's continuing
involvement - he will be back the first week in July (aimeeu,
15:58:06)
- ACTION: Bryan will
sent out email about meeting times (aimeeu,
16:00:12)
Meeting ended at 16:02:16 UTC
(full logs).
Action items
- Bryan will sent out email about meeting times
People present (lines said)
- aimeeu (68)
- bryan_att (3)
- collabot_ (3)
Generated by MeetBot 0.1.4.