Forum Discussion

RonaldMussche's avatar
RonaldMussche
Frequent Visitor
6 years ago

The exception was raised by the IDbCommand interface dedicated capacity

Hello,

Since we linked the workspace to a dedicated capacity (A2 service tier) we get this error in the powerbi service:

 

We're sorry, an error occurred during evaluation.;
Container exited unexpectedly with code 0x0000DEAD. PID: 11952.
Used features: (none). Container exited unexpectedly with code 0x0000DEAD. PID: 11952.
Used features: (none). Container exited unexpectedly with code 0x0000DEAD. PID: 11952.
Used features: (none). Container exited unexpectedly with code 0x0000DEAD. PID: 11952..
The exception was raised by the IDbCommand interface. Table: Today

 

The table is changing every time. Refresh is going fine in power BI desktop and in another workspace (not linked to a dedicated capacity)

 

We already upgraded the capacity to a A2 service tier because the memory workload was to high. But we can't find a way to see which report is using the most memory and how to reduce this. We don't have memory trashing after the upgrade.


Don't know for sure if this is related to the data refresh. The report that is failing to refresh is using OData, SQL and dataflow connections.

 

Hope someone can help us with this matter.

 

Kind regards

 

3 Replies

  • v-eachen-msft's avatar
    v-eachen-msft
    Community Support

    Hi RonaldMussche ,

     

    I found a similar issue here.

    You could refer to the following suggestions:

    The Error “Container exited unexpectedly with code 0x0000DEAD” is due to memory issues on the capacity. It could be intermittent as it is affected by other workloads on the capacity depending on what jobs are running at the time of refresh.

    To mitigate this we suggest

    1. Increase dataflows’ max memory % and decrease the Datasets max memory % by same quantity. (If they have default settings then increase dataflows to 40% and decrease Datasets to 80%)
    2. Increase Dataflows container size to 1500Mb. This will reduce parallelism and potentially reduce intermittent failures.