This an issue raised with MS, but perhaps changing the payload to encoded in the same manner SSMS and Excel do is required?
Steps taken:
New SQL 2016 instance, default instance.
Model deployed in 1200 compatability mode. Queries work fine in SSMS, model can be browsed via Excel
MSMPUMP deployed, configuration working. Can connect via SSMS with no problems.
When connecting via MSMDPUMP from a web page, queries will sometimes work, and sometimes fail with 503. Which query fails will change, but an example is attached
Once a query fails for the first time, all queries using that measure, from SSMS or Excel, will fail with
“Executing the query ...
An unexpected exception occurred.
Execution complete“
Prior to cumulative update 2, this would not only make the query fail but bring down the entire instance.
The entire request being issued is attached.
The dump files and log files are attached, as well as the two messages I see in the event viewer
There are two issues here I think:
It’s certainly a bug that the query takes down the instance, no query should take a working query from another client and break it.
Then the second is the question of whether I need to change the XMLA payload that I’m submitting
Error_Event22.txt
Error1.txt
Msmdsrv.log.zip
Request.txt
xmla4js.zip
This an issue raised with MS, but perhaps changing the payload to encoded in the same manner SSMS and Excel do is required?
Steps taken:
New SQL 2016 instance, default instance.
Model deployed in 1200 compatability mode. Queries work fine in SSMS, model can be browsed via Excel
MSMPUMP deployed, configuration working. Can connect via SSMS with no problems.
When connecting via MSMDPUMP from a web page, queries will sometimes work, and sometimes fail with 503. Which query fails will change, but an example is attached
Once a query fails for the first time, all queries using that measure, from SSMS or Excel, will fail with
“Executing the query ...
An unexpected exception occurred.
Execution complete“
Prior to cumulative update 2, this would not only make the query fail but bring down the entire instance.
The entire request being issued is attached.
The dump files and log files are attached, as well as the two messages I see in the event viewer
There are two issues here I think:
It’s certainly a bug that the query takes down the instance, no query should take a working query from another client and break it.
Then the second is the question of whether I need to change the XMLA payload that I’m submitting
Error_Event22.txt
Error1.txt
Msmdsrv.log.zip
Request.txt
xmla4js.zip