GitLab Support Operations, 16 Aug 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Support Ops - Audits - Zendesk

Description

Jason Colyer, Support Operations Manager, takes us through how to do a Zendesk Audit

A

Greetings all my name is jason collier sport operations manager here at git lab and today we're going to go over how to perform an audit funds and desk globals. Let me share my screen and we'll get started so everything's going to start with an issue in the audits project. You'll go over to your issues.

A

Click new issue, you're, going to choose the correct template in this case we're doing regular, zendesk global, make sure we apply it and the issue template itself will tell you exactly what to do from the title of the issue to assign it to yourself to assign it to complete it and all that fun, stuff, uh you'll notice here, there's the notes- and it says, enter notes here. Now it used to be the way we did. This was to manually go review everything in zendesk.

A

That's a lot to do as this company is scaled so because of that, what we ended up doing was making scripts to use instead to get this information. So here we have the zendesk audit script, I'm going to kind of go over this script and what it does and then I'll show you how to you know, run this on your local computer and then what the output looks like and how to copy and paste that into an issue and so on and so forth. So uh to get started.

A

This is using fara jet, fairing, sorry, faraday, oj, open, uri and gamble, and the ammo is going to be used to read the score team yaml file, faraday and oj to request from both zendesk and gitlab.com, oj to parse that, in a nice, easy format to use and open uri to again make another request.

A

Now what we've done here is. First, we define the client. We do this by using a new faraday connection, we're going to use, make sure it's url encoded, use the default, faraday adapter and then give it the basic authentication of the zendesk username slash token and then the actual token itself.

A

We're also going to make the request function, which is going to take http method, the endpoint any parameters we want to give into it, and then it will do a public send on that and give us the oj response, which is going to be nice, parse easy to work with now, because we're using zendesk- and we know the roles that we're actually working with we've predefined the rules to make it even easier for us, the id numbers and the role name themselves be super easy when we run the script to kind of just check with that for the git lab client we're using a new faraday connection, we're doing it with the default adapter to our encoded and all we're.

A

Sending here is the private token. In the headers of the gitlab token. Here we all function called support team yaml. All this is doing is making a request to the support team vmware file in raw format, getting it back from that and then using yaml to safe load, that for engineers we're basically just making a nice variable that we can call. That is the entirety of the support team. You have files content in yaml, parse format.

A

Now for the actual notes. Here, we've got a function that uh starts as an empty array, and it's going to essentially push the different variables different values to it. It's going to do name, comparisons, zendesk id compares group comparisons, role, comparisons and um unless the notes are empty, what it's going to do is combine them into comma separated issues if they are empty. It's just obviously going to put out an empty string in this case now the name comparison is actually pretty simple. All it's doing is saying zinda says your name. Is this?

A

The support team yaml file says this: are they the same? If they're, not it's going to report that there's a name dispatch, the zendesk id is doing the exact same thing. It's comparing the zendesk id to what we have for the support team. Yaml file groups are basically doing the same thing. What it's doing is sorting the groups from zendesk, sorting them from the support to ammo file and unless those are exactly the same array, we've got. A group mismatch roles are just like the name zendesk id.

A

It's comparing two values, one from zendesk and one from support.

A

Now what we're going to have this do? Is we're actually going to make queries to zendesk doing a search, but we're going to pull all the agents with role admin and we're just going to keep looping through this query using a new next page value to get all the advents now traditionally, it only has to run this once we don't have a lot of admins. That's by intention.

A

We don't want a bunch of advents that get hard to manage, but essentially all this is going to do is give us all the admins which we're calling users in this case.

A

So, from the next part, we're doing a loop through all the users and uh we're, including the groups on the user and we're doing a request on each user's id number. That way, we can get information about it. We're making sure that we set the users groups to the groups, the role to the actual role and you'll notice. I've kind of got something custom here, and the reason for that is there is some weirdness with administrator of it's not really a role, but it also is.

A

We also have a couple such as like myself, who is owner so they're weirdly, both an admin but not an admin.

A

um It's going through all of these users, and basically saying you know uh if they're a light agent go ahead and push the light agents if they're an admin, push the admin, if not push into agent. So what we've essentially done so far is grab every agent user of zendesk, which is a non-end user and put them into variables from this point forward, essentially, all we're doing is outputting various things based on this data.

A

Now, uh when it comes down to the admins and the agents, we're obviously we're gonna do uh we're gonna do a read on them. Do comparisons get that notes that we talked about earlier and print all that out, but that's what the rest of this is doing and then it's going to end again with more output.

A

It looks like a lot, but when we actually, when you actually see it in play, it's not, but it also is. It is a lot of output data, but the script itself is just making this easy on you now, um when you first kind of get into your you're the repo and you're on your command line, and all that there's some things you need the first to start with. Is you need ruby?

A

In this case, you know I've uh got ruby installed, I'm on 2.6.2, which, if we chat the dot ruby version file, that's exactly where we should be from there. You need the gym bundler, so I mean I can do gym lists local.

A

Bundle there's bundler right there, so cool, so what I can do is just run the bundle, and this is going to read the gym file and install all the things we need for all the audits we're doing now from here. You would uh docslash the bin folder and then you want to run the zendesk audit script, I'm not going to in this video, because the script will take a substantial amount of time easily at least 10 minutes to run.

A

It's got a lot of data to go through the easier way to look at this is to go over to this other tab. I've got and you can kind of see how it's already run. We've got a ton of data here, a ton of data information here. A lot of this is this or like all the light agents that we have and all this various information. What you would do is copy every bit of this and it's a lot you're going to copy all of this down to the very bottom line.

A

You know, you'll uh copy it and then you'll paste it into your issue. Now, uh to give you an example of what that would look like, let's, uh let's locate a past issue.

A

And here's the zendesk audit done most recently, you can see how uh this beginning part looks exactly like template, but from there we're going to have a bunch of new information that we get added all the light agents, admins and agents, but uh from the admins and agents data once you've pasted in your issue. You've submitted it. So it's all there. You can then look for any of the issues that it says need to be resolved and go through each of them ping that person via the issue and try to get this issue resolved.

A

If it's something as simple as a support team, the ammo mismatch, you see why it's mismatching, you can go ahead and just submit the merge request yourself to fix it. It's often quicker when you get to the api token part which you will have to do manually in zendesk, you will have to go grab what every api token is you're going to need to decide if this should be there or not if it can be removed, if it is we're removing it, if not, you know what is it being used for?

A

This will take time to kind of learn and the longer you work and sit down with zendesk the more you'll just know. What tokens are why they're there etc etc so feel free to reach out to your fellow support, ops, team members or myself, if you need assistance with that, but that's really all there is to doing an audit from this point forward. It's communicating the issue with the people that were mentioned in the various output and figuring out.

A

What's that what you know what to do once this is all done, you can mark it as completed close it out, and that is all there is really to it.

A

um It is a lot and again that script does take a tremendous amount of time to run, but um last I mean currently, we have something like 500, plus light agents, we're only going to keep growing, so you can imagine how manually copying just the light agents would take you a substantial amount more time than 10 to 15 minutes but uh yeah. I hope you found this educational. I look forward to seeing you in the next video.