Mirrored Oracle Error

[moritzbartram]

Hi everyone, 
I'm having trouble to create a mirrored oracle item. I've followed the set-up guide and configured my oracle database accordingly, but get the following error: 

The message appears ~30-60 seconds after selecting the oracle connection. 

Has anyone had similar experiences or a solution? 

Thank you! 

[v-kpoloju-msft]

Hi @moritzbartram,

Thank you for reaching out to the Microsoft Fabric Community Forum. Also, thanks to @tayloramy, for his inputs on this thread.

The message “A task was cancelled” usually means that Fabric was not able to finish connecting to your Oracle database while trying to fetch the list of available tables or schemas. In most cases, this happens due to a timeout or connectivity delay between the Oracle server and the on-premises data gateway, rather than an issue with credentials. Here are a few steps you can try to get this working below those are mentioned.

Check connectivity from your gateway machine: From the server where your On-premises Data Gateway is installed, try connecting to Oracle using the same connection string or run a tnsping command. This helps confirm that the gateway can reach your Oracle database without latency or firewall issues.

Ensure the gateway and Oracle setup meet Fabric’s requirements: Use the latest version of the On-premises Data Gateway (v3000.282.5 or later). Verify your Oracle version is supported and LogMiner is enabled for mirroring.
https://learn.microsoft.com/en-in/fabric/mirroring/oracle-tutorial 
https://learn.microsoft.com/en-us/fabric/mirroring/oracle-limitations 

Limit what Fabric needs to load: If your Oracle user has access to many schemas or tables, the connection step can time out. Try using a user with access to only the specific schema/tables you need to mirror.

Restart the gateway and retry: Sometimes restarting the gateway service helps clear any hung sessions or old cache.

Check for encryption or TLS issues: If your Oracle instance enforces specific encryption algorithms, make sure they align with what the gateway supports.

Hope this helps. Let us know if you have any doubts regarding this. We will be happy to help.

Thank you for using the Microsoft Fabric Community Forum.

[moritzbartram]

Hi @v-kpoloju-msft,

thanks for your reply! The gateway I'm using work fine for the connection when I try to access the data via a copy activity (within a pipeline), then I can see all my tables and preview my data. 
After inspecting the gateway logs I can see that errors appeared when I tried to set up the mirrored database, but I don't get relevant information out of it. 
The error is: 'Unable to open the Server service performance object. The first four bytes (DWORD) of the Data section contains the status code.'
The gateway version is 3000.282.5. 

I already tried restarting the gateway. Do you have any other idea? 
Thank you!

[v-kpoloju-msft]

Hi @moritzbartram,
Thank you for the in detailed follow-up question.

Since your Oracle connection works fine in the copy activity, that confirms the gateway, and credentials are good, so the issue is likely with the local performance counters on the gateway machine. The log message

“Unable to open the Server service performance object…”
is a Windows-level error that usually appears when performance counters are corrupted or not accessible to the gateway service.

You can try rebuilding them by opening Command Prompt as Administrator and running:

lodctr /R

Then restart the On-premises Data Gateway service and retry the mirror setup.

If the problem persists, make sure the Server service (LanmanServer) is running and the gateway service account has permission to read performance counters. This usually clears the “task was canceled” error when creating the mirrored Oracle item.

Hope this helps. Let us know if you have any doubts regarding this. We will be happy to help.

Thank you for using the Microsoft Fabric Community Forum.

[moritzbartram]

Hi @v-kpoloju-msft,

I ran the command, succesfully rebuilt the performance counters and restartet the gateway. Unfortunately the error persists. I updated the gateway, but it also did not work.

Any other ideas? 
Thank you!

[v-kpoloju-msft]

Hi @moritzbartram,

Thanks for the follow-up question and thanks for trying those steps and for confirming the rebuild and update. Since the error persists, let’s look at a few additional areas that can sometimes cause this issue.

Since the issue persists after rebuilding counters and updating the gateway, please double-check that your gateway is in standard mode (not personal) and registered in the same region as your Fabric workspace. Also ensure the 64-bit Oracle client version matches your database. If it is still failing, please share the recent gateway log snippet from C:\ProgramData\On-premises data gateway\Gateway*.log so we can review the next steps.

Hope this helps. Let us know if you have any doubts regarding this. We will be happy to help.

Thank you.

[moritzbartram]

Hi @v-kpoloju-msft,

thanks for your reply. The gateway is in standard mode and in the same region as my workspace. Also the oracle client version matches my database. Here is the snippet from the gateway error log:

[DM.Pipeline.Common.TracingTelemetryService] Event: FireActivityCompletedWithFailureEvent (duration=182040, err=CommunicationObjectAbortedException, rootcauseErrorEventId=0)

I hope this is helpful to you.
Thank you!

[v-kpoloju-msft]

Hi @moritzbartram
Thanks a lot for sharing that log detail that helps narrow things down.

The CommunicationObjectAbortedException message suggests that the connection between the Fabric mirroring service and your on-premises gateway was unexpectedly closed during setup.

Since your gateway, region, and Oracle client versions all look correct, this is likely a transient connectivity or service-side timeout rather than something misconfigured on your end. Please try the following next steps:

Ensure outbound HTTPS (port 443) traffic from the gateway machine is not being filtered by a proxy or firewall.

Try creating the mirror again in a different Fabric workspace or with a new gateway connection just to rule out any workspace-level issue.

If the error continues, I would recommend raising a Microsoft Fabric support ticket so the team can review backend traces and confirm if this is a Fabric-side timeout.
https://learn.microsoft.com/en-us/power-bi/support/create-support-ticket 

Thank you for using the Microsoft Fabric Community Forum.

[moritzbartram · Accepted solution]

Hi @v-kpoloju-msft,

thanks again for your reply. I've already been in contact with the fabric support team and they confirmed that the error I'm getting results from the amount of tables ~10000 that are in my database and they cannot be displayed. So this is just a UI error that can be circumvented by creating mirror items via Rest-API. 

Thanks again for your help! 

Best regards

[spaceman127]

Hi @moritzbartram ,

I ran some tests yesterday with Oracle Database version 19.3.

I also had a lot of problems at the beginning.

Maybe we can solve this.

I have the following questions.

1. Which Oracle version are you using?

2. I assume the Data Gateway is up to date?

3. Have you installed the Oracle Client and adjusted the tnsnames.ora accordingly?

4. Does your user have the permissions? Have you tested this with SQL Developer or SQL Plus?

If you have any questions, just let me know.

Best regards

[tayloramy]

Hi @moritzbartram,

That “A task was canceled” message usually means Fabric gave up while trying to enumerate your Oracle objects after the connection was picked. In practice it’s a timeout/network/gateway issue during the “Choose data” step-not a credentials problem.

Troubleshooting steps

1. Confirm prerequisites for Oracle mirroring and your gateway version. Oracle mirroring currently requires an On-premises Data Gateway (OPDG) and Microsoft documents minimum versions (e.g., 3000.282.5+) for Oracle support. Upgrade if you’re older. See Oracle limitations and the setup tutorial here.
2. Gateway health + region check. In Power BI/Fabric Admin, verify your gateway cluster is online, no high CPU/memory, and its region matches your Fabric tenant region.
3. Listener / network sanity checks. From the gateway machine, test tnsping to your Oracle service and connect with the same host/port/SID/ServiceName used in Fabric. Make sure port 1521 (or your custom) is reachable and no SSL/TLS interception is breaking the handshake.
4. Reduce what Fabric has to enumerate. Create a dedicated Oracle user with default schema and only the tables you intend to mirror (or limit permissions to a small subset). The schema browse step is what usually times out.
5. Retry after gateway restart. Restart the On-premises Data Gateway service to clear stale Oracle client state and try Create > Mirrored database > Oracle again.
6. Collect details and try again. If it still fails ~30-60s after you select the connection, capture the Activity ID (you already have it), exact UTC time, and the gateway logs, then open a Microsoft support ticket.

If you found this helpful, consider giving some Kudos. If I answered your question or solved your problem, mark this post as the solution.