GitLab Monitor Stage - Speed Runs, 27 Nov 2021

Previous Meeting

Next Meeting

⏯

youtube image

►

From YouTube: Grafana OnCall Speed Run (Part 1 of 2 )

Description

Presentation of Grafana OnCall with a feature overview and screenshots.
GitLab Issue - https://gitlab.com/gitlab-com/Product/-/issues/3380

A

Hi, I'm alana, and this is part one of two for speed- runs for grafana on-call, so this first one is all static, features and screenshots for a presentation- and I wanted to shout out to michael for attending grafana's observability con and capturing all these different screenshots.

A

My hope here is that it kind of just gives you what it gave me, which is just like a general overview of what's available in grafana on call so part two of two which will be another another separate video will be an actual dive into the product in a blank state kind of poking through on some of the features that we see here in this video.

A

So we're going to go through all of these different things, one by one and kind of point out some of the things that maybe we have or some of the things that we see that are different from gitlab's incident management solution, so integrations slack integration, alert, type rules, alert builder incidents, timelines within incidents.

A

They have alert routing and then also escalation policies and some on-call schedules.

A

So for integrations, um I'm seeing things that have a pretty wide range, so things from elastic alerts, pingdom, pagerduty and then also grafana and prometheus, and then also just a generic web hook for the slack integration they go to one channel, but within that channel it can notify specific users and, as we'll see later in escalation policies, you can specify what that channel is um in this screenshot. I also found some good ideas.

A

They have ai suggestions for on resolve unresolved incidents, so if something kind of slips through the cracks, what can that look like um to kind of resurface that and then they also have the ability which we will see in the slack integration to show similar incidents? That may have already have a postmortem great ideas, as we continue to iterate on our own solution.

A

So there are different types of alerts based off of grafana thresholds or prometheus, and when you actually get into the alert builder, there are a lot of different options to build out those alerts. So you can build it off of prometheus's query language.

A

You could also do it from custom sql within your own database, here they're showing postgresql.

A

Or from a simple math expression, so lots of different options to build out those different types of alerts.

A

Then, if we look at the actual incident itself, this is the incident list I'll show what an actual incident looks like in future screens but um kind of similar to our incident list view or comparable.

A

They have kind of an overall tree out of you and then some quick summaries up above and you can have the ability to respond to the incident from the list view so acknowledging resolving and then I believe, grafana has a feature called silencing, we'll also notice that there are multiple alerts that can be related to one specific incident.

A

Something that we're actually working on building on this milestone is incident timeline. So this is a really great validation that this is a table. Stakes feature for incidents and just looking at this screenshot in the lower right hand corner it looks like the timeline is auto-generated, and it's including information about the incident itself and escalations that took place.

A

Moving into routing and escalations um users can have personal preferences based on how they want to be notified if they are on call for a specific alert. So this includes slack sms and phone calls.

A

Something that I did notice is directly within the integrations tab. There is the ability to add an alert route and also add what the escalation chain looks like and how that alert will get escalated based on that specific integration, so taking a click, one click into that. um You can see that we selected the specific escalation path and are going to push that to a specific channel and then build out the escalation path from there.

A

So on call schedules looks like events are, who is included um to be on call. So this is a view of the on-call schedule for this specific date today um and viewing events are limited to one day at a time.

A

We have something slightly different where you can visualize for longer periods of time, but I thought it was interesting that they're limiting it to one day, something else that I noticed was um they have an integration with google calendar and we'll take a closer look at this in kind of the live demo when we get into the product and for now that is all thank you for joining, and I hope you found this helpful.

A

There is an issue linked that includes um a link directly to this deck, as well as a link to the other speedrun that I'll be doing, but I hope you enjoyed this and thank you for attending.