In the cloud environment the sqlpreview event does not fire. This is important for Linkage because it makes multiple passes to hit the style, bottomup_topdown, which deletes first, then inserts or updates next. Linkage aims to delete dependents first, then add parentage before children. If parent datawindows have both inserts and deletes Linkage sets which statement will fire and expects sqlpreview to skip others:
// Only perform the requested SQL (insert, update, delete) statements.
if (li_sqltype = CI_PreviewSelect) or &
    (li_sqltype = CI_PreviewInsert And Pos(is_UpdatesAllowed,"I")>0) or &
    (li_sqltype = CI_PreviewUpdate And Pos(is_UpdatesAllowed,"U")>0) or &
    (li_sqltype = CI_PreviewDelete And Pos(is_UpdatesAllowed,"D")>0) then
    // Allow the continuation of the SQL statement.    
else
    // Stop this SQL statement from being executed.
    // return 2 - Skip this request and execute the next request.
    return 2
end if

Instead, the INSERT executes Update(false, false). Then it makes a second pass for the Insert and gets a duplicate key.This leads me to conclude TOPDOWN (default) is safer, except for the less frequent deletes. Deletes might best work to discard child rows and cascade deletes in the database.

How do we know which buffer it will process? Is DELETE buffer always first?

Because the duplicate key happens in a PFC Linkage service and successfully updates a simple single pfc window, I set some breakpoints and use the Debug Project toolbar button of the cloud project. Rather than walking into a sqlpreview event, I get this error:

Transaction is timeout. Transaction ID: AE4651FA-79F8-4115-A3F4-28F8493C4AD3-10

No changes made to database.

This cuts short my ability to debug in the cloud environment. Timeout? Am I stepping thru debug too slow?  It's worth repeating the same code deployed as a Win32 app does not get the duplicate key error.

The console for Kestrel shows:

fail: PowerServer.Api.ServerApiController[0]
      'PowerServer.Api.ServerApiController.DataStoreUpdate (PowerServer.Api)' get result (
      {
        "IsSuccess": false,
        "RequestId": null,
        "AppName": "wastrak_cloud",
        "NameSpace": "Wastrak_cloud",
        "ErrorCode": 0,
        "ErrorMessage": "Violation of PRIMARY KEY constraint 'PK_TRANSPORTERS'. Cannot insert duplicate key in object 'dbo.TRANSPORTERS'. The duplicate key value is (XB).\r\nThe statement has been terminated."
      }).
Security Warning: The negotiated TLS 1.0 is an insecure protocol and is supported for backward compatibility only. The recommended protocol version is TLS 1.2 and later.
fail: PowerServer[0]
      Failed to commit because there is no available database connection or transaction    at ☼  .☻.MoveNext()
      --- End of stack trace from previous location ---
         at  .☻.MoveNext()
      --- End of stack trace from previous location ---
         at ♠  .☻.MoveNext()
fail: PowerServer.Api.ServerApiController[0]
      'PowerServer.Api.ServerApiController.CommitAndCreateTransaction (PowerServer.Api)' get result (
      {
        "IsSuccess": false,
        "RequestId": null,
        "AppName": "wastrak_cloud",
        "NameSpace": "Wastrak_cloud",
        "ErrorCode": 0,
        "ErrorMessage": "Failed to commit because there is no available database connection or transaction"
      }).
Since this works for a simple PFC u_dw I would rule out TLS 1.0 as a barrier on my local machine.

Chris! Thank you for your quick reply. I am installing release 1900 as I write this. Since I just downloaded PB 2022 last week, I didn't have an MR on my radar.

I looked at code xchange and I found and downloaded 2022 PFC. I'm going to inspect it.

I guess I am referring to the PowerClient App that downloads from the web server. When I run it, it looks like the interface is the same as my win32 app.

Hi Jeff;

  Can you tell us ...

1. Did you download the PB 2022 version of the PFC?
2. Have you updated the PB IDE & runtime to MR build 1900?
3. Are you referring to a PowerClient App or a PowerServer App when you say "cloud"?

Regards ... Chris

We may be having the same problem so I'll be watching this