ETL not joining despite matching columns

HowDoIDomo
HowDoIDomo Contributor
in Magic ETL

I am trying to join an ETL, but for the life of me cannot get the join to work. I have double and triple checked that the columns are formatted the same, trimmed on both sides just in case, etc and Magic ETL still won't recognize that the data matches. What should I do?

Tagged:

Best Answer

  • DavidChurchman
    DavidChurchman Coach
    Answer ✓

    The preview is limited to the first certain number of rows of data, so if there are no matches at the top of the inputs, it might not find any data to preview after the join, just give you a green check if it's a logically sound join.

    Please 💡/💖/👍/😊 this post if you read it and found it helpful.

    Please accept the answer if it solved your problem.

Answers

  • EvilGenius
    EvilGenius Member

    Can you share the screenshot of the Magic ETL & the data from both datasets?

  • MarkSnodgrass
    MarkSnodgrass Coach

    Joins are also case-sensitive. I often will convert text to uppercase to use for joining and then drop those columns later.

    **Check out my Domo Tips & Tricks Videos

    **Make sure to <3 any users posts that helped you.
    **Please mark as accepted the ones who solved your issue.
  • DavidChurchman
    DavidChurchman Coach

    Have you identified that there are matching rows for each side of the join? Sometimes when troubleshooting a frustrating ETL, it's good to start with a very small subset of your full data. Here, you could filter both sides of your join for a value (or handful of values) that you believe should match, create an output for both before the join, and and an output after the join and inspect what's happening.

    Please 💡/💖/👍/😊 this post if you read it and found it helpful.

    Please accept the answer if it solved your problem.

  • HowDoIDomo
    HowDoIDomo Contributor
    edited July 18

    @EvilGenius @MarkSnodgrass @DavidChurchman

    Thank you all for replying. I think I found a fix for my issue.

    I was trying to match on ID numbers where one dataset treated them as text 00011111111 vs. an integer 11,111,111. So, I tried to normalize the match columns to both be text using data handling and drop the zeroes using a "split column" tile:

    ETL_FAIL.png

    That didn't work. So, I added string operations to trim both sides of each column. That didn't work either. I'd still end up with no matched data even if I got both IDs to follow the format of Text 11111111.

    Today I ran into the same dilemma working with on another file. This time instead of switching from Integer to text, I switched the text column to an integer. It dropped the 000 but didn't add the commas like the other file. So, I cast the integer column 11,111,111 as an integer which dropped the commas and allowed it to match.

    ETL_Worked.png

    The weird thing is. The preview didn't show any data even though it ultimately worked.

  • DavidChurchman
    DavidChurchman Coach
    Answer ✓

    The preview is limited to the first certain number of rows of data, so if there are no matches at the top of the inputs, it might not find any data to preview after the join, just give you a green check if it's a logically sound join.

    Please 💡/💖/👍/😊 this post if you read it and found it helpful.

    Please accept the answer if it solved your problem.

Categories