What were you trying to do that didn't work?

I'm trying to start colocated resources all at the same time, even with batch-limit of 300 I see that schedulerd is scheduling actions at the same time but execd isn't executing start actions at the same time

What is the impact of this issue to you?

For application to work correctly we need db2_partition resources to start all at the same time because there is dependency between db2_partition resources. What I see is that when we put high load pacemaker is running start and waits until start action finishes ( either success or failure ) only then starts other resources in the colocation set

If this is something that happens due to constraint can you suggest a way to guarantee that Pacemaker will run actions in parallel?

Please provide the package NVR for which the bug is seen:

How reproducible is this bug?:

Easy when load on system is high

Steps to reproduce

1. Apply high load to system
2. Have multiple resources colocated together with infinity
3. Set batch-limit to very high value ( 300 used in this case )
4. kill node where resources was running on
5. observe that pacemaker will move resource one by one instead of starting at the same time

Expected results

I expect resources to get started in parallel

Actual results

Actual result showed that resources are started in sequential fashion