GitLab Pipeline Execution, 15 Dec 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: GitLab 13.8 Kickoff - Verify:Continuous Integration

Description

CORRECTION: Mentioned in the video but NOT planned for 13.8 is "API to bulk delete artifacts" (https://gitlab.com/gitlab-org/gitlab/-/issues/223793)

A

Welcome to our uh presentation for what we're planning to work on in 13.8 and we're going to give you a sneak preview into the designs of features that we are planning for our one year focus within our three-year vision. My name is tao yeager. I am the product manager here at gitlab, for continuous integration and um joining me is my designer extraordinaire I'll. Let her introduce herself hi, I'm vithika, I'm the product designer for verify ci.

A

um So what I'm sharing with you right now is the direction page for continuous integration, and I'm just going to talk through some of the things that we have described here under what's next and why you probably already know, if you've seen our direction page, that we've been working to make it easier for you to run a pipeline with preview pre-filled variables on the run pipeline form that is actually going to be releasing later this month.

A

That should make it easier for you to manually trigger a pipeline without having to manually, enter all the key value pairs for the variables of your pipeline.

A

Next, we're in the process of working on a way for you to turn off, keep latest artifact um by implementing the ability to configure this at the project level.

A

This is currently in progress with the target of releasing in 13.8 and then a fast follow to that will begin work in giving you that same ability to configure at the instance level for self-managed instance and also not mentioned here, is the ability to at the group level configure whether or not you want to keep latest artifact um and then finally, thinking even beyond that, we want to go back to something that I had mentioned in a previous kickoff.

A

Video is giving you the ability to an api really endpoint to let you quickly delete artifacts in bulk again that all of those features should give you more control over the artifacts that are being kept uh in your repository uh and actually allows you better control over your storage uh consumption.

A

So what I want to really turn most of our attention to in this session is the forward-thinking designs we have in our one-year focus for for continuous integration. We want to empower users with actionable insights, about pipelines for the very important purpose of letting them uh effortlessly, monitor, debug and optimize their pipelines. So um I'm not gonna read through all those uh three bullet points, but um vitica is going to walk through the ways that we are going to implement.

A

um What's described here in these three bullet points, ways to make it easier for you to react to pipelines that are running to optimize pipelines based on analysis of historical pipeline runs and then also improving the law, job log experience, so that it helps with your debugging efforts, I'm going to stop sharing and let vitika take it from here.

A

B

Much char um yeah, I would start talking about the designs that we're working on and um from the previous slide the tao uh had very recently shown. I would only be focusing on the first and the third point and just remind you of what they were about. So the first point talked about uh accessing insightful information about the running pipeline and the third was uh I mean allowing for easier debugging. I mean improving the whole experience around uh debugging jobs, so uh yeah.

B

So what you're seeing here is the new revamped design uh layout for the pipelines overview page, and I would quickly talk about some uh very major changes that we have made here and the reason that we have done so so um I mean all the information that you see today on the pi brands overview page. They have been very thoughtfully uh condensed into this corner section, and we have also introduced something. That's sort of a pie chart a small donut chart that talks about uh the status of the jobs which are included in the pipeline.

B

So this gives a very quick glimpse of. What's going on inside the moment. You enter this page now. um The next thing you'll notice here is this primary action button and just to like mention that this primary button is going to be a dynamic one, it would take into account the status of the pipeline and what is the most important action in that in the context of whatever is going on with the pipeline at that particular moment.

B

So this will uh keep changing depending on that, and of course, this is something that you might not have noticed on this page, which is a huge graph that we are showing, and there are many information that we are trying to uh expose divulge using this craft. So you see these pills uh here on the graph uh on the scan chart and in the in these pills, there's a division of color. So um the gray part here it talks about for how long this job has been cued for.

B

However, this uh other colored section this uh talks about so, for example, this is red in color that denotes that this pipeline it ran for this duration, and then it failed. Similarly, this next one, it kind of uh it had a shorter run than the previous one and it succeeded and so on, and then the next thing you will be seeing on this graph is the small, tiny warning icon and what it does uh is something I'll explain in the upcoming slides.

B

So this is what the graphic would look like when we are maybe at a very initial stage of the pipeline, when there are just about three or four five jobs running and as the pipeline progresses. This is going to get more and more populated, and now um I mean users might want to have a quick idea about how many jobs are running or how many jobs are have failed.

B

So we are providing that information using this tooltip and similarly, we'll have a tooltip for this warning icon here, and this also gives kind of some very important information about that particular job. So here it talks about the duration for which this job was running, then, for how long was it cued, and it also provides a very important information, which is how many jobs does this particular job triggers? So this gives a very concrete idea about like how many jobs uh which are to be run are depending on this one particular job.

B

So um I mean it's just helping users to identify the major bottlenecks in their pipelines and similarly, if you see that when I hover over this, then this one is also getting highlight so there's a very uh clear connection between this graph and the pipeline graph that we have at the bottom so, and the next thing is uh something that we would probably be working.

B

I mean uh we have worked on it but would be implemented once we are done with all the changes on the pipeline graph- and that's also, I mean something worth showing, so I would just go ahead.

B

So when I click on this job here I mean right now, I'm just hovering, but when I click on it, what we do is we bring the job logs on the same page as the pipeline overview page. So now users would not be going to a whole different page when they want to view the pipeline job blocks and we are also working on organizing the logs better so that it's easier to identify like what the problems are with my uh job. Looking at the logs, for example, these are the collapsible sections that you're providing.

B

So all the jobs that have been completed succeeded not they get collapsed and there's a very clear indication about their status using a colored border. So, for example, if I want to know about this one, the second job that ran and I'm just curious on looking at certain details about it, so I just click on it and here it uh it's exposed, and similarly I can just click again and collapse it, and then we are also thinking of providing a set of actions which are associated with that job.

B

Using this menu a drop down menu- and I mean clicking anywhere outside this panel- it would close it. And yes, that's pretty much all uh that. I wanted to talk about today. Tao back to you.

A

um I think, on the job logs, that open from the right panel should have a ta-da.

A

Sound added to it, I think it's great because previously and currently you have to navigate away from the pipeline overview page and then navigate back. If you want to see the progress of the pipeline, so anyways super excited folks for these design concepts that we'll be focused on in the coming year. 2021 is going to be exciting if 2020 was not exciting for everyone. Anyways thanks for listening, have a great holiday and.