GitLab Support Operations, 19 Oct 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Support Ops - Zendesk - ZD-SFDC US Federal Sync

Description

Jason Colyer, Support Operations Manager, takes us through how the ZD-SFDC US Federal sync works

A

Hey y'all uh jason collier support operations manager here, I'm joined by james and caleb uh today, we're gonna go over the zendesk salesforce us federal, sync scripts, the code and how all that jazz works. So let me share my screen and we'll kind of get started.

A

Right on uh so right now we're looking at is the v4 branch hasn't quite been merged into the main master branch. Yet, but this is the new version of the code and uh yeah right now. We've got our readme file, it kind of tells you like the required gems, which needs to be updated, the file tree path and then the soqo queries which we'll go more over those when we actually go to the scripts, but uh the big meat and potatoes of how everything works.

A

At least sync wise is the ci cd script uh right now it's using ruby, 2.6 later version. I'm gonna update that to like three or whatever, but the stages we basically have are the gather, compare sync and then we're syncing users as well, because federal actually does contact syncing a much smaller pool of users and orgs were able to actually do that. um There's like a future iteration where that won't be a thing, but we'll kind of cross that bridge when we get there.

A

um We have a cast path to vendor ruby. It's just to make the stages go a little faster between because they don't need to re-download. You know every gem every time they just use the cache for the previous ones.

A

Our before script is pretty simple. It's going to echo out, you know it's going to print out the ruby version and then install bundler and then run bundle. Some people take different approaches. I like this approach, just because it's consistent between scripts of yeah, I'm pretty much always going to just install bundler and then run bundle and then it'll have a gem file that will tell it what to actually install.

A

We have this job, that's the precinct, zd org report, which is me trying to shorten a potentially very long name. But essentially this is a gather data stage. It's only running via schedule.

B

A

It's running the gathers and desk organization scripts, which we'll go over in a bit. It's going to create an artifact that lasts four hours and it is going to be named zendesk forge.json, not original naming so that's just kind of what we go with uh the precinct, sf board report or precinct salesforce organization report. It's doing much the same thing: it's gathering salesforce accounts.

A

This one, however, creates two different artifacts we've got the salesforce accounts which we'll actually be using for the sync, and then we have this account problems json, which we can download later, and this will tell us what problems are going to occur, what are being skipped in the sink um when we go over the actual salesforce stuff, we'll kind of go over, that um we got the zd user report, which is doing the same thing. Just gathering, zendesk users, salesforce user report again same thing, just gathering salesforce contacts.

A

We have the precinct comparison for orgs, which is going to compare them and create a creates and updates file to let the sync know this is what to create. This is what to update. um I use needs here to basically say I need these precinct zd or report and precinct sf org report jobs to succeed. Without those we can't proceed because it needs the artifacts they create the user comparison, it's kind of the same thing, but for users, and then we have the sync organization and sync users job these require.

A

The comparison report is completed since it needs these create, update files, and all this is going to do is run sync organizations or sync users, and that's what's actually going to update orgs or users in zendesk or create them cool.

A

So beyond that, uh here's our gem file right now not a lot here, because we basically just need json, oj and restforce a fun fact about that. You can click these and it'll actually take you to the rubygems since that's the source.

A

That's always like that feature in gitlab, um but basically because oj is going to require faraday, because that's how oj works, I don't have to say I need faraday, because it's gonna already be installed with oj, but basically, when we run bundle, it's just gonna install these gems and um json is a library. It's not it's, it's a gym, but also it's not a gym. It's weird and then rest forces salesforce's api gym.

A

So um our scripts here um are pretty much. How a lot of this is running and I'll show you just one of them, because they're all gonna follow the same format so like gathers in dashboards, it's going to require the zdsfdc sync us federal file, and then it's just going to call zendesk salesforce sync, us federal, zendesk gather organizations run they're all going to follow this format. They're pretty much all going to be calling different libraries that we've built and doing the run function, so the more important stuff would be what's in lib.

A

This is our main file. This is what we're actually requiring, and all this is is declaring the module zendesk salesforce. Sync us federal. I really need to shorten that name somehow um and it's requiring bundler setup and then doing a bundler required default. This is fancy talk of hey, so, whatever I said, I needed in the gym file go ahead and tell the script that I require those gems so that they're already loaded and ready to go and then from there it's going to require the client file.

A

So the client file we'll go over that cool, so the client file again we declare our module and then we're making a class called client. It's going to require the compare, salesforce and zendesk files which we'll go over in a bit.

A

I set up a hash of retry options here and basically we're going to pass this into faraday and it's going to tell it how to handle failures and most common failure. Being I timed out, I lost the socket. I dropped a packet, nothing ruins the script more than I dropped a packet, so I stopped running completely. So this is just a way of saying: hey!

A

uh Do five max retries wait a second between each one, um there's an interval randomness of 0.5, which kind of it multiplies by the interval to kind of give it a range. So in this case point half a second to a second is what it's going to weight.

A

um The back off factor is two: that's if it gets an error that tells it hey, you've been api, limited uh back off factor tells it okay, wait this much time back off. Factor in this case is two seconds.

A

I've never actually used that one in the scripting, but it's a good thing to put in place um and then we've got exceptions, and this is where it actually will use all this stuff. So exceptions for this would be connection failed. Timeout, that's the exceptions. It's actually looking for to use these retry options um beyond that we're making a salesforce object at salesforce makes it kind of a global variable if you've never seen this symbol before double pipe equal. It's a really fancy ruby term for set it to this.

A

But if it's already set to something, don't do anything, it's a really good way to be able to call this function again without it retrying to make everything in the function instead. It'll just say: oh, you just want the salesforce object.

A

In this case, I'm calling the restforce and giving in all the required variables that it needs to actually generate a salesforce connection, zendesk kind of the same thing, but instead of using a gym, I'm actually using faraday. um I haven't found a zendesk api gym that I really like um there's a couple. I've played with they're okay, but they kind of fall short in some areas, especially on retries. That's where they really fall short on.

A

So I ended up just going pure faraday on this one um and for the faraday connection, you know we're giving it the zendesk federal, url, the api url and then we're going to have it make sure the requests retry knows to use these retry options we have up here.

A

The adapter is whatever the faraday default. Adapter is, for you know the server. Whatever it's running on for headers, we set the content type to application json.

A

This isn't required for get requests to the zendesk api, but any post or put request is absolutely going to require that because it will only take json objects and then we're using basic auth, which at some point in the future, I will update, because faraday is now deprecating basic auth to use a different method, but uh for now basically I use my username or the sync's username slash token, and then the api token.

A

And then here I have requests uh this pairs with the zendesk object and it's basically calling the public send function, which is a nice wrapper for uh faraday, has a bunch of like request, dot, get request, dot, post, request, output, request, dot delete um instead of having to worry about that and remembering the exact right function to use uh public send will instead take an http method, the url and parameters and then just use it that way.

A

So I can call you know, request git, you know google.com and it'll do a get request at google.com instead of me having to remember, is it get or does it get send or whatever function? I, like public, send it's a cool wrapper, but we're passing all that into a response and then we're using oj, which is a nice gem that will take an http response and make it into a ruby object. So it's nice and easy to work with in this case, I'm telling oj to load the response body.

A

So that's the client file um by itself. It doesn't do a whole lot. It's gonna be calls to it. That will actually do something.

A

So uh we'll go with salesforce as a starting point. We have the salesforce rb, which you saw the client file require. All this is doing is requiring the gather script, and that's just so. I could kind of separate stuff out.

A

So if I go into salesforce here's the gather script, all this is doing is requiring accounts and contacts. um I like to do it this way, just to make it a little easier to read. If I I mean you could put all of this in one file, but it gets kind of hard to read after a while, so uh the accounts you know, we've got our module, our salesforce class, our gather class and then our accounts class, and I'm telling it to inherit from client.

A

The reason for this is, I don't want to have to call zendesk salesforce sync us federal colon, colon client, dot salesforce every time I need to call salesforce by doing it. This way, I'm basically telling it just just inherit from it know that salesforce means the one from the client.

A

Here's our run function, which is uh what our scripts our bend scripts actually use. But you can see how it's doing gather verify report and artifacts gather is essentially just going to print out that it's gathering the data it's going to call the data function and then say it's done.

A

The data function um see how I've got a variable data that I'm setting as a global with the at sign using this double pipe equal and then I'm using the salesforce gym to run a query and it's using a query string and then I'm taking the results and mapping them to account objects.

A

So the query string. This is our query string for accounts if you're not familiar with sql or soql, which is the salesforce object, query language. um Basically, what we're doing here is I'm telling it what I want.

A

So in this case I'm saying yeah, I want the account id 18c, which is fancy api talk for the 18 digit account id, and then I want the name, and I want car this account c, which is the api for the arr for the entire account, instead of just specific subscriptions, which is what we use in our sinks: ultimate parent sales segment, employee c, which is really fancy for.

A

What's the market segmentation, large smb pub sac um account owner calc c, that's uh in salesforce account owner is an object and calc c is basically a fancy way of saying right, so just calculate the account object and give me to the account owner and give me that person.

A

This is going to end up returning a string that will then sync over number of licenses the number of seats in their license, their total licenses, the type of account which would be like customer former customer prospects, all that fun stuff technical account manager name, which is the name of the the tam support level, is the support level um that's determined based on subscriptions, but we actually have the script doing a little more with it.

A

We'll go over that a bit um region which we actually don't use in us federal, but I wanted them kind of in sync with what the other scripts are doing. So I called it anyway um gs health score color, which is the gain sight, health score color.

A

um I don't understand fully how this health score is determined, but my understanding is gainsight is using a bunch of information about like potential potential to renew um emotional state stuff, like our non-emotional state, you know, like health state of the account stuff like that to determine a red, yellow, green color system or null, which means it hasn't calculated anything.

A

Some people find that useful. So we do sync it over uh next renewal date is going to look at their subscriptions and look at the next upcoming renewal and put that there uh solutions, architect, lookup dot name. uh Basically, I'm saying do account solutions, architect, lookup on the account and then give me the name of the person which we'll call down below this is one of the newer parts of it right here. I'm calling a sub query and I'm basically doing I want the current subscription status the end date.

A

The start date and the plan name from the subscriptions that are active for the accounts that we're pulling. This will allow us in the script later to actually look at the subscriptions and then ensure that the support level matches what it should be. It'll also allow us to add tags in case we have customers who are I'm ultimate in gold and bronze and starter.

A

I've got all these plans, but we want all those tags present and then, as we work tickets, we can remove the ones not applicable, uh so we're adding we're asking for all that from the accounts table or accounts object. I should say since soql's objects and then we've got our where statement, which is uh basically us being specific for us federal. So we only want customer former customer prospects yeah. I know that's kind of everything.

A

Everything but leads um the account. Terry should account. Territory should start with u.s pub, fed uh percentage signs of glob, just like in sql, glob being fancy term for a wild card and then we're doing basic. We're basically doing this right here, which is using the plan. So if their support level is premium ultimate or expired, they're good, we want them if they're basic. We also require that their lice number of licenses is over, is equal to or over 2 000..

A

So this kind of huge query here is basically just saying: yeah give us all the us federal accounts. That's that's the summary of what this says.

A

So the salesforce salesforce is going to run that query and then map it into an account object which down here you'll see. This is just turning all of those into a giant hash that will be passed along into an array um and basically, what you're seeing here is I've got some stuff like type is down case, and then I wanted to take any spaces and make them underscores that way. It follows camel casing for health score.

A

I want it to a string and down case to a string is nice, because if it's null it'll make it an empty string, so it doesn't error out. It's just a kind of nice way to do that, but uh you can see how some of these we have. uh We have. This check account function. That starts first, so check account is one of the newer stuff. I've got this plan called nine expired.

A

They all have numbers for a reason. It's just for sorting of what plan is the most important.

A

Basically we're going to go through the account subscriptions um if you're not familiar with this right here, the um ampersand.each it's a really nice way of. If the subscriptions were null and I called it dot each it's going to say I can't iterate on a null. It should have been an empty array. I don't know what to do. I'm going to error out center, etc, ampersand's, basically saying if it exists and each so it's a nice little trick. Just so you uh don't have to add a function and you know add a call there.

A

That's, like you know, return blah if subscriptions null no like should say so. It's a nice little trick that I've picked up along the way um from there. I've got kind of a redundant line here that says, skip it. If the subscription status isn't active. The reason that's redundant is, I already said, only pull ones where it's active, but you know it's there. I I don't know I'll remove it at some point.

A

um If the current ending is nil, we just want to skip the subscription, because that basically means it's an invalid in date that something's wrong with it. We don't want to parse that um if the date parse of the end date is less than today, that means they're expired, so we're going to go ahead and just skip the iteration.

A

Otherwise we're going to uh determine plan and push that to our plans. Array up here and all determine plans is. Is we've got one ultimate two premium, three starter three starter, and this is basically just doing a little regex on the name of the plan, so ultimate becomes ultimate premium becomes premium, starter, starter and basic is also starter.

A

um The numeric values will come into play up here. We've also got eight custom, that's just an else statement. Basically, if it didn't match any of these, just call it custom. It's gonna look way. You know that way. We know something's up, but basically, um if the plans are greater than one meaning, we've added something other than this nine expired we're going to delete expired.

A

That means they have some kind of plan from there we're going to do a sort and map, and uh this right here this one dot dot negative one is a really fancy way of saying yeah. Give me the string minus the first character, so basically ultimate or one ultimate becomes ultimate to premium becomes premium, and I'm calling unique that's if they have like five silver subscriptions. We obviously don't want silver silver service over. We just want one silver, because yeah um be beyond that.

A

I say I want the minimum plan and uh that's one: that's why these have these numbers just for sake of sorting of one two, three, eight nine, if you're wondering why it skips from three to eight to nine global, has more plans and had to account for things like community and silver and bronze and gold.

A

But basically all this is going to the entire point of this check. Account is to set our subscriptions and set the support level to what it actually should be the highest supported plan that they have and from there we're going to pass it into the account object.

A

So beyond that we got this verified data which, in our case, just checks for duplicates. That's the only thing we actually have to check, for we can sync the rest of it just as is, and all that's going to do is duplicates um up here. I've got this fancy. I've got this where duplicates is equal to determine.

B

A

Okay, so all I'm doing is taking the data and I'm grouping it by the name and I'm down casing it, because zendesk cares about names regardless of case, so we want all the names in down case, so we can compare them and I just want the values. What this is basically going to do is give me a whole bunch of account objects, that's what it's going to do it, but it's going to group them in arrays based on their name and then from there.

A

I want to select any of this grouped value where the size is greater than one, and I want to flatten it, which is basically going to say, move it out of an array all right, because select's going to put it in arrays and then I want to take it out of that and just have that.

A

So what this basically is doing is, if you have two accounts with the same name, they're going to be listed in this duplicates and then from there you can see how we do minus equals, so I'm removing all duplicates from the data, because we can't sync duplicates there's no way for the script to know which duplicates the right one.

A

So we just remove them and report on them from there. You know we've got a report which will tell us hey, I located this many here's the duplicates and then we'll also iterate and tell us each of the names of the duplicates and it's going to create our salesforce accounts and sftc account problems, json artifact files.

A

So I know that was a lot any questions before I proceed.

A

Right on so, the other gather script for salesforce is contacts. um You'll see this looks a lot like the other one our run gathers and generates, and it reports um a lot of this is really the same. We've got, you know determining duplicates. The only addition here is the missing values, because if they're missing an email or a name, we obviously can't sync them. We don't yeah. Those are required fields a desk, but it's basically just going to do the same thing. Here's our soql query, you'll notice, the where statement pretty.

B

A

Exact same with the addition of, if it's an inactive contact, we don't want that. So this is kind of a weird statement, but an active content is false, which would make them an active content. Contact. Sorry, hey jason! I'm sorry! um When you asked for questions, I hadn't seen that uh artie had posted something in chat. Are you meant vocalizing for the recording.

B

Oh yeah, no, um I think jason kind of answered this question when he was talking through it already, but um it's just to do with the way that the plan is set um like for the tagging for the different plans, and I assume that the reason why there's only the three plans like ultimate uh premium and starters, because this is u.s fed and then global- has all of them.

A

Correct yeah global also has a bronze silver gold community, so uh it does have a couple. Others that's why it goes from one two three to eight like it's literally, because this is kind of a mirror copy of what we have in global, which I I think does like one ultimate, two gold, three premium, four silver, something like that. I don't remember the exact numerical, um the exact order of it but yeah it has like community. I think it has like edu oss. It has a couple different plans that we have to account for.

B

A

Cool uh yeah, so you'll notice the contact sync, the contact gathers pretty much the exact same thing other than we're gathering contacts instead of organizations.

A

Our contact object is considerably simpler, because all we really need to know is game, email, org and we're going to use the salesforce id just for sake of data checking later so it's there, but uh you'll notice here, like if the names nil or the name is blank or the email is no or the email is blank. It's basically adding these to a missing array that way it can kind of subtract them.

A

From our contacts to say, I can't sync someone without a name, I can't sync someone without an email duplicate same thing: we can't sync two different. You know two of the same email to two different users.

A

Cool, so that's the salesforce portion of the sync script, um the zendesk function, the zendesk portion yeah, it's going to look kind of like the salesforce function. You know salesforce stuff to begin with, it's going to call gather. It also has sync which we'll go over in a bit: we'll focus on gather.

A

Okay, cool, so organizations, um you know sales, zendesk, salesforce, sync, us federal class, zendesk class gather class organization inherit from client. That's just so. We can call the requests really quickly without having to type out a ton of space. um What this does is just generate.

A

The artifact and report um probably could have made a function called gather, or you know, probably could have put like a gather function here, but I just didn't for this: one um data is gather, orgs gather, words is going to basically say I'm gathering the organizations it's going to set an empty array have next page set to one and we're going to do what might be like nails on a chalkboard to some programmers, but basically I'm going to iterate over pagination using faraday requests.

A

um I have not found a better way to do this in zendesk, yet I always look forward to finding a better way, but right now the most they'll return is 100 results page. So the second you're over a hundred organizations surprised we are for every instance, um you're gonna have to start doing page nation. So basically this is going to print a dot for each page. This is just so when I review scripts.

A

I can see how many you know pages it had to go through um it's going to make her get requests to organizations json setting the page to whatever next page is so start with one it'll. Add the organizations to our orgs array up here. It's going to make sure to map them into org objects. We'll go over that in a second.

A

um If the next page is blank in the response like it doesn't have a value. That means we're done paginating we can just break out of it. Otherwise we need to add one and go to the next page after that it puts out done and returns the organizations, so the org object, you'll notice. This looks a lot like what salesforce had that's, because we're going to compare them in the compare script, so it helps to have them look almost the exact same.

A

There is a slight difference in that there is the tags, we're pulling tags from this, because we want to make sure the tags that we you know we want to make sure we add all the tags that we need for them.

A

um We also make sure to check the support level and determine the subscription so for support level. Basically, I'm just checking to if it says it's basic. I want to say it's starter, because basic's, an old term, we don't use that starter is the correct term. Otherwise, just return the support level that it has invest for the subscriptions.

A

I've got an allowed the one for global way longer, because it has a lot more different things to check for, but in the case of in the case of the us federal, we got starter premium ultimate community and custom and community actually doesn't belong there. So I mean it doesn't hurt that it's there, but there's no.

B

A

In federal, but basically we're going to duplicate the tags um by just using org tags dupe or dup. However, you want to pronounce that I've always pronounced it dupe since it stands for duplicate um and then we're gonna set that to subscriptions and then we're going to only select from the tags, the ones in this and these select allowed tags these plans, um if it didn't find any it's gonna return expired because they don't have any plan tags. That means they're expired and then it's gonna return subscriptions.

A

If you know it's gotten that far, what this is going to do is basically duplicate all that um subscription checking we did in the salesforce stuff and then uh clean tags. This is literally just cleaning up the tags so that we don't have a bunch of duplicates and we don't have the starter. You know we don't have these plan tags in there, since those will be added in later, but uh that's pretty much all there is to gathering zendesk organizations and um surprise users is kind of the same stuff.

A

There we go, uh you know, gathering users, it's gonna page eight over uh the only difference here is we're using the user's endpoint um I'm, including.

B

A

This means I'm sideloading the organizations, because otherwise what I get is an organization id which is not as useful to me since what I have to pass to zindesk to create or update the user is not the id. I need the name of the organization.

A

So what I do is sideload organizations, and this will basically give me my object- will be like users and then organizations- and I can just cross reference between the two. So I have the name. I don't have to make two different api requests for it. I instead just have one.

A

um I only want end users like we don't need to sync users, or you know, agents and then page next page. It's kind.

B

A

Doing the same stuff we saw before you know we're taking the user object, we want the! U and we want all the organizations passed in each user object function. This is so we can easily cross reference them and it'll do that until the pages there are no more pages and then return the users, so uh the user objects we're going to um detect from the orgs where the user.

A

You know the organization id equals the user's organization, id that'll, give us the org and then down here we can call you know if the org is nil, just make it empty they're not associated, um it's not gonna happen very often, since they would have to be associated to have been detected anyway, but um you know, and then the org name will be returned if it exists. uh Salesforce id is kind of the same thing of like. If there's not a no an org, it's gonna be blank.

A

Otherwise past the organization's cust organization field, salesforce id.

A

Cool, so the next part would be gather but before compare, but before I go over that any questions so far.

A

Cool then I'll go over the compare again like these, these rb files that are in the parent folder. Here they don't do a lot other than just require the actual individual files.

A

But we've got compare and organizations.

A

Cool, so we have our uh modules and salesforce sync us federal. That's the last time I'm going to read that entire thing out.

A

Then we have our compare class and our organizations class and we've got our run, which is going to do, compare generate artifacts and report.

A

So uh we're setting some variables here, just for sake of referencing them later it's good for creating the artifacts, it's good for doing quick counts, but our actual compare is going to determine our sync needs. It's going to go over each of the accounts, which are defined way down here, should have put that at the top. But that's basically the salesforce accounts from our previous artifact file and it's going to go over each with index.

A

Each index is useful for basically just printing out this nice pretty formatted string. So we can see what count it is on. um That's really not needed other than that, but I like it so yeah, but all this prep format doing is it'll print out a count of what account number it's on, so that we can quickly when we're running the script. See like. Oh I'm on five of five thousand. I'm on two thousand eight hundred of ninety two thousand- um it's nice to see how long it's running it's nice to see where it's at.

A

um It's also really nice if something errors and breaks the script, because I have a nice count of where it stopped.

A

um But beyond that we're going to look in the organizations which again don't find down here, but it's the zendesk organizations um and we're doing a detect to look for a matching sales force id since that value is only going to be present in one organization or should only be present in one organization.

A

um That's a good thing to look for if it's not in zendesk, which would mean that zd is nil. We know that we need to create an organization, so we're going to push it to creates using create objects.

A

um Otherwise, we need to determine if updates are needed, which is you know, we got a and zd uh the account in the zendesk object and then we'll print done if you're not familiar with backslash r, it's returned so it puts the cursor all the way back to the beginning of the line which is useful when I'm printing out this stuff and just overwriting numbers and stuff like that.

A

So um we'll look at the create object, real quick! This is a create objects. um You'll notice. This looks different than the objects we had in previous scripts. That's because this is specifically what zendesk wants it to look like when you pass it to it. It's for a created, a non-existent org. It wants a name.

B

A

Organization fields, which are our custom fields that we've made into zendesk- and you know, salesforce id, sfdc, short id support, level, arr, market segment, etc, etc, etc, and then tags we're basically getting the subscriptions and just sorting them. That's it because the rest of these, like market segment or support level or um account type, those are drop, downs they'll, add their own tags automatically to zendesk. So we don't actually have to worry about that.

A

We just have to make sure we add those subscription tags that way if their support level is ultimate, but they also have premium and starter, they will get ultimate premium and starter tags in their organization.

A

Now uh the determine update function, um that's what we got here. We've got a comparable object, which is a very fancy way of saying. I don't need to compare an id or the tags. Those are custom only to zendesk. I do not need to compare those to salesforce, it's making sure to sort the subscriptions and make sure the health score is in string format. This is just to make sure that when it compares, it doesn't feel like well, it said premium ultimate starter, but zinda said starter premium ultimate.

A

If it does that it's going to be like these aren't the same thing. They mostly are.

A

But yeah, so basically, it's just going to duplicate the organic organ. Remove those pretty up the objects so that we can actually compare it, which is as equal as easy, as is the account equal to the zendesk object. Are they the same thing? Well, if they are, then we're going to push it to do nothing just for sake of counting and reporting, otherwise we're going to push the update object, which looks very similar to the create object other than it has an id beyond that. You know, update tag list which you can see here.

A

All that's doing is making sure to add the subscriptions, making sure to sort them and unique them and then pass them back. uh Then we generate the file report. Is you know, words create sync and good standing? That's where do nothing comes into play? It's just. This is how many orgs are good, um not necessary. I like it, though just so I can quickly look and make sure the sync is working, it's good to see the sync run and then on the next run, see that number go up like we want to see the do.

A

Nothing count. Go up and up and up yeah, so uh the other compare is the users.

A

Unsurprisingly, it's going to look very very similar other than the create and update objects are going to be different, like that's as simple as they are that they're not complex, they don't have to do tags or subscriptions they're, just users they're a lot easier to deal with.

A

So any questions on compare before I go to sync.

A

Cool so syncs our last portion of this, um which is basically us taking that creates and those updates files and passing them to zendesk to actually do it. um We have these nice little objects here that create updated errors. They're just nice placeholders, while we add stuff to them um creates, is here. I think I probably define them at the bottom yeah I need to stop doing that, but creates is the creates file updates. The updates file for the creates I'm going to do each with index.

A

That's just for sake of reporting, create org is just going to call create a post request to the organizations using create objects, which I then take the object we had before and really says to really put it into the object that zendesk wants, because they want it in the format of organization and then all your fields and then organization fields, and all that um I do it this way, because zendesk has occasionally changed how they want those objects to look, and it is really annoying to have to figure out what to have to go back to compare and then go back to other another another another.

A

It's a lot easier just to take what we already have and at this end stage say just put it in this format: that's what they want.

A

um After that, the uh you know we'll create the org, um the sync orgs, which I should probably rename to update words. It's badly named um we're doing something a little different, you'll notice, I'm doing it each slice 100..

A

That's really fancy way of saying, take my giant array and make it into smaller arrays that are that have a size of 100., um basically from there we're going to map them into update objects, which you can see up here, um we're then going to take the requests and put them. You know and you'll notice that the actual sync object I'm using is organizations and then a bunch of these update objects and it's putting that to json and sending it to the update mini.

A

The reason for this is if we have to update 100 orgs, doesn't sound like a lot, but it takes 100 api requests. Update mini takes one. It creates a job which we'll have the job status for, and the job process is way quicker than the api. Does. I don't know why it just does we don't do that for creates solely because I've not seen a case where I had to create more than like 10 use. 10 works as long as the script stays update and everything's good to go.

A

It really doesn't have that situation and then, at the end, you know it's going to report. Here's what I created. Here's what I say if there were any errors during these scripts, which uh you can kind of see here, there's the handle error during the create org. If there was a response, error, it'll print out all the various errors which most the time is name blah already exists, I've never seen it send out anything else, because our sync script has so much data checking in it that it's prevented every other possible error.

A

But uh I've absolutely seen name already exists and that's normally when ops has to go investigate. What's going on and the create use. You know the sync users is the exact same thing: it's just using specifically the user create the user, create many and all that fun stuff.

A

um There's not a whole lot of difference on this one. Either, like you can see how it's still doing each slice, it's doing a user's update. Many, the only difference is these objects are way smaller and a lot easier to manage.

A

Yeah. That's uh pretty much all there is to that. So um any questions on the sync portion or any part of the script.

A

Cool looks like we're in a good place, so uh yeah, that's the basics of the zendesk sales force, u.s federal, sink, which I said it again so I lied earlier. I guess I had to say it one more time, but uh that's the basics of it, um the code it lives in the ops instance. That's because the pipelines they they're gonna show customer data it's private and protected.

A

We can't have that available to everybody, so unfortunately we have to have any private instance, but um we do have our training and our documentation all that fun stuff. That is publicly available for you to look at this video as well. um You can always feel free to come in to support ops and ask us questions. You can always feel free to schedule a coffee chat with me if you want to go over it or understand or if you have ideas to improve it.

A

Yeah with that, I hope you found this educational. I look forward to seeing you in the future.