Scheduled refresh error DM_GWPipeline_Gateway_TimeoutError

[Nameless_LW]

Our customer has an issue with a dataset scheduled refresh. Problem started on 15. March (14. March update went normally, nothing that I know of has changed in the customer environment or in the report itself.

When the dataset tries to refresh, after just about 1 - 1.5 minutes it always fails with this kind of error:

--
There was an error when processing the data in the dataset.
Data source error: {"error":{"code":"DM_GWPipeline_Gateway_TimeoutError","pbi.error":{"code":"DM_GWPipeline_Gateway_TimeoutError","parameters":{},"details":[],"exceptionCulprit":1}}} Table: NNNNNNNNNNNN.

Cluster URI: WABI-WEST-EUROPE-B-PRIMARY-redirect.analysis.windows.net
Activity ID: 08d5fa8e-5dd5-45ed-a51a-6d2b2362717c
Request ID: 191471c1-13ce-4010-98d3-8d9a45b44046
Time: 2023-03-27 10:33:52Z
--

The table name in the message seems to change between errors, so the issue doesn't seem to relate a specific dataset table. The error also occurres in manual refresh (on demand), but only sometimes (I haven't figured out any pattern).

Report is using a On-Premise Gateway (standard mode) to refresh data from an on-premise SQL Server. The Gateway has been updated to it's latest version (3000.166.8) and restarted. The timeout setting in the report has now been set to 120, but it didn't have any effect.
The Dataset is quite large, about 700Mb and it's not using incremental refresh. Despite the fact that dataset is large and supposed to be a bit slow to refresh, I think it's weird that the timeout happens so fast (under 2 minutes).

Gateway logs don't seem to tell anything. Comparing the logs from the 14. March (when everything was still ok) to later days show that when the refresh fails, nothing seem to prevent the refresh from happening, it just stops doing anything.
The lines in the logs seem very similiar (between successful and failed refresh), but after about 80 lines of doing things, the Gateway just stops and then reports the "Pool cleaner connections removed: 0, count: 0, buckets: 0" line continously, like when idle. No errors or other unusual things are reported to the logs. QueryStartReport or QueryExecutionReport -logs don't have anything about the failed refreshes, like there wasn't any queries to execute at all.

The server (where the Gateway has been installed) is acting normally and doesn't seem to have issues with inadequate resources.

There have been same kinds of problems reported here lately, is this related to them? If so, do you have a guess, when the issue might be resolved?

[v-yetao1-msft]

Hi @Nameless_LW 

Is the workspace where the dataset is located shared capacity or Premium capacity ?

Best Regards,
Community Support Team _ Ailsa Tao

[Nameless_LW]

Hi @v-yetao1-msft 

The workspace is using Premium capacity, the dataset isn't using the large dataset format (I think at that size, it shouldn't need it. and refresh has been working without any problems before).

Br.

Nameless_LW

[jpaolino]

We were having the same issues where all of our dataset refreshes were failing with timeout errors.  This was on the March 2023 release of the standard gateway version.  

We rolled back to Feb 2023 (3000.162.10) version and all of our reports have been refreshing normally this morning. 

[Nameless_LW]

This problem got fixed on it's own a week and half ago. Nothing was changed, the scheduled updates just started to go normally without any problems. No news from MS if they have done something at their end.

[jpaolino]

@Nameless_LW , good to here.  I wonder if your gatweay got the April update and corrected the issue.