System Created Backup Tables

[alozovoy]

Occasionally I end up with a system created backup table in my Fabric Lakehouse. For example, if my table is "customers", I also see a table named "customers_backup51c54568_b46b_40f9_8db2_800abb09622f". It appears to be a one-time backup. It does not get updated after it is initially created, even if the actual table does.

This happens more frequently to the larger tables in my lakehouse (10,000,000+ rows). It also sees to happen to tables imported to the lakehouse using a Copy pipeline activity from a SQL server.

When I see these backup tables, I manually delete them.

Does anyone else see this happening in their lakehouse? Is there a reason why these get created?

[jpelham]

This has recently started happening in one of my lakehouses in the past week. I'm now up to 6 extra days worth of tables. I cheked the enable staging setting in the copy data activity and the box is not checked, so I don't think that this should be occurring. Like the poster, these are tables being loaded from a SQL server.

[ilseeb]

hello, I started facing the same issue.

Did you find a solution?

[jpelham]

At first I was just deleting all of the tables that contained the word backup in their name using a notebook. I would run this every few days rather than every day. In my situation this was occuring a lakehouse and pipeline where we were doing testing, so it was not critical to keep it running. I ended up setting up a new pipeline and lakehouse and have not had any further issues. 

If deleting the lakehouse is not an option, I would suggest trying to replace the pipeline with a new version. If that is not an option, running a notebook to delete the backup tables is going to be the easiest way to mass delete them. 

[ilseeb]

Hi, I used @string('OverwriteSchema') instead of Overwrite and this still worked and the backup tables are not being created anymore 🙂

[jpelham]

Thanks for the suggestion. That is one thing we tried, but in our case it didn't resolve the issue. I am glad it worked for you! Since I deleted that lakehouse and created a new pipeline, we have not had the issue return in any of our lakehouses. Hopefully it stays that way 🙂

[v-shex-msft · Accepted solution]

HI @alozovoy,

I thin they are staging tables use to copied from the source data store to the staging storage.

You can take a look at the following document to know more about this feature and how it works:

Copy activity performance optimization features - Azure Data Factory & Azure Synapse | Microsoft Lea...

Regards,

Xiaoxin Sheng

[Apelsinas]

The OverwriteSchema instead of Overwrite solved this for me. I was using just 'Overwrite' in my metadata table before, then I checked the JSON code of the pipeline and there it was the overwrite option described as 'OverwriteSchema'.