cancel
Showing results for 
Search instead for 
Did you mean: 

SAP Datasphere - dpAgent seems to be unstable

Mandar_Damle
Explorer
0 Kudos

Hello Experts,

We are using dpAgent for remote tables with real-time access. I observed that the delta suddenly stopped working for one of the flows though it is active in the data integration monitor. For the other two flows, real-time access is not getting enabled. The job runs for 3 hours resulting in a time-out error attached below.

Currently, our datacenter team is looking into the dpAgent logs but I feel that the functionality of a remote table with real-time access is too fragile compared to replication flows. Overall, dpAgent is not a very stable component in the Datasphere landscape. Can experts throw some light on this?

Thanks!

Regards,

Mandar Damle

Accepted Solutions (0)

Answers (1)

Answers (1)

thorebedey
Advisor
Advisor
0 Kudos

Hi mandardamle ,

the DPAgent is indeed a component that can be fragile.

Due to insufficient sizing or wrong usage it can crash. From seeing your screenshto it could be that your initial load is too big because it takes 3 hours. You can decrease the load by partitioning or decreasing the data amount.

You are correct, repilcation flows are more resilient and are also the integration option that SAP will focus on in the future. So if possible, integrate data through replciation flows.

BR,
Thore

Mandar_Damle
Explorer
0 Kudos

Thank you, Thore, for responding!

I was trying to pull less than 5K records, but it couldn't handle that much volume either. The datacenter team couldn't find any specific reason to understand the root cause. In our case, the size is 8GB for QA and 16GB for Prod but I'm not sure if these numbers are any help. Is there any document about the sizing of the DPAgent?

I was able to enable the real-time delta after a few days, but I had to do it again from scratch.

  1. Remove the old snapshot
  2. Re-load a new snapshot
  3. Enable real-time again

So, it does not make sense to use it in a real production scenario as there are millions/billions of sales order/purchase order records.

Anyway, as you have suggested, I would try to use replication flow wherever a high volume is expected.

Thanks again!

Regards,

Mandar