Forum Discussion
Materialized Lake View - 'Invalid Date, Invalid Date'
- 11 months ago
Hi TheFifthJoin ,
Thank you for providing the details. Even reverting to a previous version may not fully resolve it, since cached metadata can persist across sessions and environments. anilgavhane , suggestions are still relevant and worth trying.
For the"Private link check failed"This is probably not directly related to the MLV issue, but may be impacted by the same OPDG update. This error often occurs if
1. The Fabric workspace or capacity isn’t properly registered in the Private DNS zone
2. The Private Link endpoint is misconfigured or not fully propagated
3. You’re using On Premises Data Gateway, which doesn’t fully support Private Link
If you’re in a secure VNet environment, you might consider using the VNet Data Gateway, as it’s designed for Private Link and can help avoid these errors.
Alternatively, you can
- Disable redirection temporarily and test creating the MLV with a new name and explicit casting
- Check Private DNS zone records to ensure Fabric endpoints are properly registered
- If using OPDG with Private Link, try switching to VNet Data Gateway
Helpful Reference:
Solved: Seeking Advice: Private Link and VNet Data Gateway... - Microsoft Fabric CommunityThanks for prompt response anilgavhane .
Regards,
Yugandhar. - 10 months ago
To update - bit of a mixed bag of solutions.. The rollback of the OPDG seemed to fix the columns going into strange orders when we added/removed a column from the materialized lake view. I've not seen this as a known issue, but rolling back seemed to resolve and we'll wait for a new update release before trying to update the OPDG again.
The 'Invalid Date, Invalid Date' seems to be being displayed when my date value in the raw data is NULL. If we load this via the sql endpoint in SSMS, the fields display as null values as expected.. a bit of an odd display, but I believe its just a display quirk, rather than an actual issue with the data. Hopefully this will get addressed in a future release also, but for now, ignoring it is doing no harm as the 'Invalid Date, Invalid Date' message doesnt pull into powerBI visualisation either.
Finally the Private link issue was a much wider issue that affected all of south-west connections. Something to do with an error in an update MS pushed - they address, rolled back internally, and the error disappeared.
Thanks for checking back!
Subject: Re: Materialized Lake View - 'Invalid Date, Invalid Date' Issue After August OPDG Update
Hi there,
This issue appears to be a bug related to schema drift and column mapping introduced in the August release of the On-Premises Data Gateway (OPDG).
When you add or drop a column (like dropping age), the gateway or engine seems to misalign the underlying column positions or data types. As a result, non-date string columns (like title) get misparsed using a date parser, outputting 'Invalid Date, Invalid Date'.
Here are a few troubleshooting steps and workarounds you can try to resolve this:
Force a Full Refresh / Schema Sync:
Try triggering a full refresh or manually saving/re-publishing the Materialized Lake View query. This often forces the gateway to rebuild the metadata schema from scratch instead of relying on cached column indexes.
Re-create or Update the Definition:
If a refresh doesn't fix it, briefly alter the view definition (e.g., re-add and re-remove a dummy column or recreate the view) so the engine updates its target schema mapping.
Check Connector / Driver Types:
Ensure that the connection string or driver used by the updated OPDG hasn't defaulted to an explicit date/time parsing setting for text fields.
Rollback OPDG (Temporary Fix):
If this is blocking production work, temporarily rolling back the OPDG to the previous month's version (July release) should restore normal behavior until a patch is released.
Report to Microsoft / Product Team:
Since this started right after the August OPDG update, it is very likely a regression bug in the gateway's connector layer. I recommend opening a support ticket with Microsoft and attaching your gateway logs so their engineering team can release a hotfix.
Hope this helps! Let us know if a full refresh fixes the mapping on