Project

General

Profile

Actions

action #81106

closed

coordination #69310: [epic] SUSE QA tools team ticket process helpers

test out chat service notifications, e.g. matrix, from github actions size:M

Added by okurz almost 4 years ago. Updated almost 2 years ago.

Status:
Rejected
Priority:
Low
Assignee:
Target version:
Start date:
2020-12-16
Due date:
% Done:

0%

Estimated time:

Description

Motivation

Based on #69310 we find useful limits, SLOs, queries to check but we are unsure about notifications, e.g. how would we react to GHA pipeline failures. Maybe it helps to receive slack notifications

Acceptance criteria

  • AC1: GHA pipeline failures in os-autoinst/scripts trigger matrix notifications
  • AC2: team members are aware of where the notifications come from and what needs to be done
  • AC3: We are not annoyed by too many frickin' unactionable alerts

Suggestions

For slack one can
try out
Slack Messaging
in https://github.com/os-autoinst/scripts GHA pipelines

Don't spam channels but definitely not community channels :)


Related issues 2 (1 open1 closed)

Related to openQA Infrastructure (public) - action #77317: chat bot to conduct daily checks, alerts, reminders, etc.New2020-11-10

Actions
Related to QA (public) - action #102059: Integrate the Slack feed notifications feature for progress queriesResolvedokurz2021-11-082021-11-19

Actions
Actions #1

Updated by okurz almost 4 years ago

  • Tracker changed from coordination to action
Actions #2

Updated by okurz almost 4 years ago

  • Project changed from openQA Project (public) to QA (public)
  • Category deleted (Feature requests)
Actions #3

Updated by okurz almost 4 years ago

  • Related to action #77317: chat bot to conduct daily checks, alerts, reminders, etc. added
Actions #4

Updated by okurz over 3 years ago

  • Priority changed from Normal to Low
Actions #5

Updated by okurz over 3 years ago

  • Target version changed from Ready to future

this was an idea by the team and for the team but apparently there is not much interest so I will move it out of the backlog for now

Actions #6

Updated by okurz over 3 years ago

  • Subject changed from test out rocket chat notifications from github actions to test out chat service notifications, e.g. rocket chat, from github actions
Actions #7

Updated by livdywan over 3 years ago

  • Subject changed from test out chat service notifications, e.g. rocket chat, from github actions to test out chat service notifications, e.g. slack, from github actions
  • Description updated (diff)
Actions #8

Updated by okurz over 3 years ago

  • Subject changed from test out chat service notifications, e.g. slack, from github actions to test out chat service notifications, e.g. matrix, from github actions
  • Description updated (diff)
  • Target version changed from future to Ready

I am not really motivated to support proprietary tools before free software so I suggest to use matrix/element first

Actions #9

Updated by VANASTASIADIS over 3 years ago

  • Assignee set to VANASTASIADIS
Actions #10

Updated by VANASTASIADIS over 3 years ago

  • Status changed from Workable to In Progress
Actions #11

Updated by VANASTASIADIS over 3 years ago

There is a working matrix solution (currently on a personal test repo, but easily reproducible in any workflow with minor additions). I am wondering however:

  • currently in the scripts repo there are 2 jobs, containing 3 steps in the workflow: JOB 1: a) check WIP limits b) set due dates JOB 2: c) run ci tests on push

I'm wondering: since the first job is a scheduled job, should a notification be sent in every failure? That's certainly easier (and less complex), but if for example the job fails on a Friday evening, by monday we'd have a lot of spam in the chat.

Another solution would be saving the previous condition in a file in the repo, and comparing to see if anything changed. That would lead to notifications only the first time something fails. But it would still miss other cases: for example, a case where the same job/step fails but for a different reason.

I think it's simpler in the case of https://github.com/os-autoinst/qa-tools-backlog-assistant: notify only on status change for every query. If "overall" backlog is off limits, you get one notification: the next one will be when it's inside limits again.

Depending on the job in question, different ways of notifying may be prefferable: for some jobs only once on every change, on others only one report in fixed intervals... I would appreciate other opinions and thoughts here.

Actions #12

Updated by okurz over 3 years ago

  • Subject changed from test out chat service notifications, e.g. matrix, from github actions to test out chat service notifications, e.g. matrix, from github actions size:M
  • Description updated (diff)
Actions #13

Updated by livdywan over 3 years ago

VANASTASIADIS wrote:

There is a working matrix solution (currently on a personal test repo, but easily reproducible in any workflow with minor additions). I am wondering however:

What solution is that? Do you have a proof of concept implementing this?

  • currently in the scripts repo there are 2 jobs, containing 3 steps in the workflow: JOB 1: a) check WIP limits b) set due dates JOB 2: c) run ci tests on push

I'm wondering: since the first job is a scheduled job, should a notification be sent in every failure? That's certainly easier (and less complex), but if for example the job fails on a Friday evening, by monday we'd have a lot of spam in the chat.

What do you consider "a lot"? I would suggest to aim for one notification a day. I don't care if we see the exact same the day after, that just means we need to catch up.

Actions #15

Updated by VANASTASIADIS about 3 years ago

  • Status changed from In Progress to Feedback
Actions #16

Updated by livdywan about 3 years ago

We still need these items here:

  • We need #suse-qe-tools to receive notifications
  • Add secrets to the GitHub pipeline
  • We need a bot account for Matrix
Actions #17

Updated by okurz about 3 years ago

  • Related to action #102059: Integrate the Slack feed notifications feature for progress queries added
Actions #18

Updated by okurz about 3 years ago

@VANASTASIADIS what are your own plans on this? Should we unassign you and pick it up within SUSE QE Tools?

Actions #19

Updated by VANASTASIADIS about 3 years ago

  • Assignee deleted (VANASTASIADIS)

@okurz So, for matrix this should be working, as long as someone adds MATRIX_ACCESS_TOKEN and MATRIX_ROOM_ID to the secrets. I don't have the permissions to add secrets, so someone with the appropriate permissions should do that.

In addition, I see that this has rolled back to being slack-centric. So I will unassign and feel free to assign and add the secrets, or proceed with a slack implementation. If it's not too urgent and you have no free hands, I can tackle the slack implementation too, after I'm done with a couple of qe-core tickets. Ping me in that case.

Actions #20

Updated by okurz almost 3 years ago

  • Status changed from Feedback to New

Actually hasn't rolled back to being "slack-centric". For Slack I was merely using quite a different feature in the related but not same ticket #102059

Actions #21

Updated by okurz almost 3 years ago

  • Status changed from New to Workable
Actions #22

Updated by mkittler almost 3 years ago

It is still not clear to me whether this should now be implemented for Matrix or for Slack.

Considering https://github.com/os-autoinst/scripts/pull/108 it seems the Matrix part might have already been concluded and only Slack is left. Even if that's the case I'm still wondering where the Matrix messages end up as the room name it literally a secret (and one can only update/remove them).

Actions #23

Updated by okurz almost 3 years ago

  • Target version changed from Ready to future

mkittler wrote:

It is still not clear to me whether this should now be implemented for Matrix or for Slack.

Considering https://github.com/os-autoinst/scripts/pull/108 it seems the Matrix part might have already been concluded and only Slack is left. Even if that's the case I'm still wondering where the Matrix messages end up as the room name it literally a secret (and one can only update/remove them).

Yes, the matrix part is only done when we actually see messages, not in before :) The priority should be 1. Matrix, then 2. Slack. But as this apparently wasn't interesting for the team to pick up for some time I will for now remove the ticket from our backlog again until someone or something convinces me again :)

Actions #24

Updated by okurz almost 2 years ago

  • Status changed from Workable to Rejected
  • Assignee set to okurz

okurz wrote:

I am not really motivated to support proprietary tools before free software so I suggest to use matrix/element first

I give up ¯(°_o)/¯ #102059 must suffice for now

Actions

Also available in: Atom PDF