Are there pre-configured templates for IT Asset Management in JSM?

Hi Gabriela,
Thanks for the very comprehensive and precise answer.

I was also looking for a template with existing data (around 200, 700, or 1,000 records) where I can run all the necessary tests, queries, etc.

Thank you again for your time and assistance.
Have a great day!

Hi @Giuseppe Miccoli, Atlassian doesn't publish one. There's no schema file anywhere to load either, because Cloud exports objects per object type as CSV and nothing above that, which is what JSDCLOUD-9967 is asking for, Under Consideration on 584 votes. Copy Assets data does move whole schemas with their objects between sites, but it clones a site you already have, so it won't seed a new one.

What Atlassian does publish is the column shape, in the ITAM import tutorial. models.csv is Model Name, Category, Unit Price, Description. assets.csv is Device Type, Serial Number, Model Name, Purchase Date, Owner Group, Status, Asset Tag, PO Number, Asset Status. Fifteen rows each.

For your 200, 700 and 1,000 runs, put those columns in a Google Sheet and fill down, then point the import at the sheet instead of uploading a file. Schema settings, Import tab, Create import, CSV Import, then Import data from a web address. The sheet has to be public and on https. Swap /edit#gid=0 on the URL for /export?format=csv and turn on "Viewers and commenters can see the option to download, print, and copy". After that you change the row count in the sheet and re-run the same import to step the volume.

Leave Automatically create object types and attributes switched off, which is what the tutorial itself does. Auto-create gives you one object type named after the file, with text attributes taken from the header row, so the model column arrives as plain text with nothing to traverse. Map it as a reference yourself and put Name = ${Model Name} in the AQL field on the assets mapping. Run the models import first so those objects exist when it goes looking for them.

Thanks.

Hi there,
Thanks for your reply and for taking the time to share your thoughts.
For this specific case, however, I am looking for a slightly different solution.

Best,

The sample on that tutorial page will not seed a 1,000-row test, and it is smaller than the page itself claims. The prose says 40 objects, ten of each category. The data block underneath it is 15 rows: four phones, four laptops, three servers, four printers. assets.csv is 15 too, SN001 to SN015. So there is no real-data file to find, and that is the honest end of that search.

Generating one gets you closer than hunting for one anyway, because your object types are not the tutorial's. This writes both files at whatever volume you want, in the exact column shape the tutorial documents:

import csv, random
from datetime import date, timedelta
N = 1000
CATS = {"Laptop": ("Dell XPS","HP EliteBook","Lenovo ThinkPad"), "Phone": ("iPhone","Galaxy S","Pixel"),
"Server": ("Dell PowerEdge","HP ProLiant"), "Printer": ("HP LaserJet","Brother MFC")}
models = [{"Model Name": f"{v} {i}", "Category": c, "Unit Price": random.randint(200, 9000),
"Description": f"{v} {i} {c.lower()}"} for c, vs in CATS.items() for v in vs for i in range(1, 6)]
with open("models.csv", "w", newline="") as f:
w = csv.DictWriter(f, ["Model Name","Category","Unit Price","Description"]); w.writeheader(); w.writerows(models)
with open("assets.csv", "w", newline="") as f:
w = csv.DictWriter(f, ["Device Type","Serial Number","Model Name","Purchase Date","Owner Group","Status","Asset Tag","PO Number","Asset Status"])
w.writeheader()
for i in range(1, N + 1):
m = random.choice(models)
w.writerow({"Device Type": m["Category"], "Serial Number": f"SN{i:05d}", "Model Name": m["Model Name"],
"Purchase Date": (date(2019,1,1) + timedelta(days=random.randint(0,2200))).strftime("%d/%m/%Y"),
"Owner Group": random.choice(["IT","HR","Sales","Finance"]), "Status": random.choice(["Active","Inactive"]),
"Asset Tag": f"AT{i:05d}", "PO Number": f"PO{i:05d}", "Asset Status": random.choice(["In Use","In Stock","Disposed"])})

I ran that at N=1000 and it gives 60 models against 1,000 assets with no orphan references. That last part is the bit that matters more than the volume: every asset row points at a Model Name that exists in models.csv, so the reference mapping resolves. Random model strings import fine and then leave you with a thousand unlinked objects.

The dates that script writes are day-first, and the tutorial's own samples are ambiguous the same way, so pin Date Format under More options on the import rather than letting it guess. The doc is upfront that it guesses when you do not: "Assets will automatically try to find the correct format if none is provided". On a thousand rows where most days are under 13, a wrong guess is silent.

Change N for your 200 and 700 runs and re-run the same two import structures.