There was a problem refreshing. Cannot acquire lock for model because it is currently in use

[dbeavon3]

My dataflows are acting more buggy than normal.

I've been trying to refresh a stand-alone dataflow all day long.  It always says:

There was a problem refreshing your dataflow. {"RootActivityId":"d68e57a4-07ec-3fd4-d8f3-2e94c4e7xxxx" "ErrorMessage":"Cannot acquire lock for model '861d221d-e2fe-43a4-a01e-8be2df05xxxx/24884407-c942-43ff-a6a4-eb372af6xxxx' because it is currently in use."}

This is a standard gen 1 dataflow in premium capacity, pointed at a web/REST data source.

Any ideas what this means?
I also hoped to get around the problem by renaming the dataflow and uploading a "fresh" one to replace it.

However that doesn't work for me either.

I'm opening a support ticket (probably my tenth in the past month).  This Power BI team must REALLY love hearing from their disrupted customers!  At this point I think they added all these bugs in there so that they can talk to me once a day.  I'm not sure why they want to talk to me so much... I'm not actually that interesting to talk to. 

[Original-IceMan]

Had a similar issue last Monday and called Micorsoft Support and initally were told wait 24 hours and it will clear. Our report refreshes every two minutes for a reason and u want us to wait 24 hours????  Inbetween the time we submitted our ticket and when MS got to us we had built a duplicate dataflow and moved connnections. A week later after some prodding was told that it can be resolved sooner if needed by calling Microsoft, however no contact information or instructions were given on how to do that if it was needed again. Thanks for helping Microsoft out with there support.

[dbeavon3]

I have a more information about these locking issues with gen1-dataflows in the PBI service.

The back-end error is CDSALockAcquireError.

The most relevant information that was shared with me is :

"Occasionally, a dataflow will fail to release a lock, usually due to a refresh failure or a user cancellation. Most of the time, these issues auto-resolve after 24 hours due to the 24 hour limit on refresh completion. However, sometimes the 24 hour rule fails to take effect, or the customer does not want to wait 24 hours. 

If customers don't want to wait a day, there are live-site engineers that can be contacted at Microsoft (by CSS Mindtree/professional) who have a special procedure they can use to release the dataflow locks"

That part above is very important to learn, as a PBI developer.  Whenever I'm told by a company manager that they need their data urgently, then they probably wouldn't find it acceptable if I tell them to chill, have a beer and come back in the morning for their data.  Managers would hold their report-developers responsible for getting their reports working, and would probably believe we had initiated these problem in the first place...   In this particular case the dataflow-locking issues that we encountered were NOT actually triggered by me.  ... there were other unrelated problem in the health of Power BI that triggered these dataflows to get all locked up.

Unfortunately for the PBI developer, there aren't any easy workarounds either.  This is because we can't delete or rename a locked dataflow, and lots of other PBI artifacts are referring back to the original dataflow.

Additional second-hand information (use at your own discretion):

"Premium dataflow read and write locks are very normal. It is how dependent dataflows (linked entities) are kept in sync with one another, similar to how databases use transactions. Refreshing a dataflow and/or saving a dataflow is designed to fail when locks are active, with the following user error:

Cannot acquire lock for model '[workspace id]/[dataflow id]' because it is currently in use.

The following is normal, expected dataflow behavior:

• Locking ALWAYS happens at the Dataflow level, not the entity level.
• Within a single workspace, when a refresh is running, all downstream dependent dataflows are locked during the refresh and show as refreshing themselves in Power BI.
• Attempting to edit and save any of the refreshing dataflows results in the expected "Cannot acquire lock" user error.
• Attempting to cancel an upstream refresh will cancel that and all dependent refreshes. The upstream refresh will show as "Cancelled" in Power BI, but the downstream refreshes will show as "Failed". 
  (see images)

• Within a single workspace, refreshing a downstream dataflow Does NOT cause the upstream dataflow to automatically refresh.
• Attempting to edit and save OR refresh the upstream dataflow while the downstream dataflow is running will result in the expected "Cannot acquire lock" user error.
• When there are two or more upstream dataflows (A1, A2) that a downstream dataflow (B) depends on (linked entity), if A1 and A2 are refreshed at the same time, one will acquire the lock and the other will fail.
• Across workspaces, a downstream or upstream dataflow can be edited and saved OR refreshed regardless of the status of the dataflow in the other workspace. Downstream dataflow will use the latest complete snapshot from the upstream dataflow.

Occasionally, a dataflow will fail to release a lock, usually due to a refresh failure or a user cancellation. Most of the time, these issues auto-resolve after 24 hours due to the 24 hour limit on refresh completion.

However, sometimes the 24 hour rule fails to take effect, or the customer does not want to wait 24 hours. 

Causes of Locking Issues

One or more of the following may be observed to indicate the presence of locks -

1. Dataflow refresh is STILL RUNNING as observed in Power BI and taking longer than normal.
2. Dataflow refresh is STILL RUNNING as observed in Power BI after 24 hours.
3. Dataflow refresh is STILL RUNNING as observed in Power BI and clicking on Cancel refresh results in a "Something went wrong" error in the Power BI UI.
4. Dataflow edit and save fails with "Cannot acquire lock" user error while the current, upstream, and downstream dataflow refreshes are NOT ongoing."

Conclusion 

Hopefully this is helpful.  Ideally Microsoft would share this type of thing first-hand in their "known issues", rather than making customers regurgitate it ourselves in the community.  I am assured that none of this is overly sensitive and I received confirmation from my source that nothing here is a "secret".  It can be shared with friends and relatives and coworkers... so feel free to put it in your next Christmas letter!

This information was provided in July 2024 and is subject to change if/when Microsoft fixes the locking bugs.

[v-yajiewan-msft]

Hi @dbeavon3 ,

Thanks for letting us know your user experience.

According to your case description, I do understand how frustrated you are now.

Thank you for sharing this information for those on the forum who are experiencing similar issues.

Best regards,

Joyce

[dbeavon3]

@v-yajiewan-msft  Hopefully it is obvious that it should be Microsoft who is sharing what they know about these bugs, and not customers.  Can you please post your internal wiki and KB and tell us what you learn about this new error (CDSALockAcquireError)?  It appears to be new to Power BI this year.  I never encountered it in the past, and nor do other customers.  My fear is that Microsoft is either aware of the issue and unable/unwilling to fix it... or that they haven't actually worked on this one yet, nor tried to create a repro of their own.

I don't want to spend time fighting some mysterious internal locks in the PBI service which we have no control over.  In this case the issue came up after another series of bugs in the service this past week.

Questions, based on what you find in your internal wiki/KB:

- How often is this error/bug reported by your customers?

- When did customers start encountering it for the first time?  In 2024?

- What is the length of time before the lock (CDSALockAcquireError) is internally released?

- What is the goal of these locks?  Why do they persist long after a dataflow has been cancelled?

- What types of PBI artifacts are affected by this bug?  Is it just gen1 dataflows or other artifacts as well?

- Has Microsoft already deployed any software changes (fixes or workarounds) to mitigate the problem?

Any information would be helpful.  The CSS organization was somewhat familiar but hasn't been encouraged to share any meaningful amount of information yet, other than to tell customers that we must wait 24 hours for the problem to go away on its own.  That is a pretty unacceptable "solution", for most customers.  Admittedly we haven't engaged the PTA/FTE yet on the Microsoft side, and I'm hoping there will be a little more transparency once Microsoft has awareness of our case as well as Mindtree.

[dbeavon3]

According to a support ticket I created, this painful issue is actually by design.

Microsoft appears to be placing some sort of internal "locks" on dataflow components when a concurrency/synchronization issue takes place.  The locks will prevent your environment from working properly for at least a day, and causes a situation to go from bad to worse.

According to the Microsoft engineer, I'm told that:
"A lock was automatically placed on the dataflow. This lock should be removed after 24 hours automatically, and the refreshes show work as usual."

.. I'm still trying to find some public documentation, since none of this makes any sense.

[dbeavon3]

This bug appears to be another well-known bug.  Here are some other users impacted by it.
https://community.fabric.microsoft.com/t5/Service/quot-CDSALockAcquireError-quot-when-attempting-to-...

The error message is associated with the code, CDSALockAcquireError.