Unexpected behaviour when scheduling a flow containing other flows

[FredrikHansson]

New day, new scheduling problems!

We had some difficulties with scheduling in Viya before. The first (Jobs and Flows - How do I schedule a flow?) turned out to be a bug, the other (How do I make my scheduled flow to actually run?) a user mistake.

Let´s what´s the problem this time. 😊

The task

We have three different flows “Personal-data”, “Senioralert 1-8” and “Swedeheart 1-8”. Each of the flows contains a number of SAS Studio flows who are to be run in a certain order.

“Senioralert 1-8” and “Swedeheart 1-8” can run in parallel, but “Personal-data” must run befor them both.

We created the flow “Nattbatchen” ( “The night batch” in swedish 😊) that contains “Personal-data”, “Senioralert 1-8” and “Swedeheart 1-8”.

💡 The flows in the pictures below are built in Environment Manager --> "Jobs and Flows". Not to be confused with the kind of flows that are built in SAS studio.

Here are pictures of all four flows:

(as you can see there is more happening in “Nattbatchen”, but it shouldn’t be relevant to my issue)

My problem

When I trigger the main flow “Nattbatchen” manually (by right-clicking and selecting “Execute” everything executes in the order I expect it to.

But when I schedule “Nattbatchen” the order in the subflows ar not respected. All SAS Studio Flows in the “Personal-data”-flow are triggered at the same time.

I hope these pictures will describe the problem better:

Expected behavior (happens when running “Nattbatchen” manually)

As you can see the pdbatch*-flows run one at a time. Sa2batch*-flows run one at a time, shtbatch*-flows run one at a time (in parallel with sa2batch* which is what I want)

Unexpected behavior (happens when “Nattbatchen” is scheduled to run at a specific time)

All pdbatch-flows are started at the same time. Shtbatch* and sa2batch* waits for pdbatch, but are then started all at the same time.

We are using Viya 2022.09. Is there someone experiencing the same problems?

[FredrikHansson]

The problem has now been reproduced in another environment. So at least I'm not alone. 😉

A support track is opened: 7613725502