SOLVED SCCM Orchestration Groups: Order of orchestration

[angrygoat]

I've created an orchestration group to patch SQL Clusters. The pre- and post-scripts trigger scheduled tasks that run under the cluster administrator account to move cluster roles away from nodes starting orchestration.
I've run the orchestration 4 times (no manual triggering, starting upon entering the maintenance windows). The first and the third time, it all worked fine but the second and fourth time the orchestration seemed to have issues.
My Cluster has two nodes, node1 and node2. The Orchestration Group is configured to run in sequence, starting with node1.
Node1 runs it's pre script (triggering a scheduled task and moving the roles away), installs the pending updates and scheduled a reboot.
And in that state, node2 ran its pre script, moving the cluster role back to the node that is about to reboot.

How is that possible?
As I understand, in sequence mode (or any other mode of orchestration), a node should run the pre script, run the pending deployments, reboot if necessary and after the reboot run the post script. After the post script on a node has run, the next node in the sequence acquired the orchestration lock and starts with the pre script.

Thanks for your help

 

[angrygoat]

Just realized this issue is addressed in MECM 2603. I will run a site upgrade and re-test.

 

[angrygoat]

I've ran Site Upgrade to 2026-03 and rebuilt my orchestration group - unfortunately the Issue persisted.
The first server runs the pre script, installs updates and schedules a reboot
The Reboot delay is 30 minutes as defined in the client settings.
At this stage, everything looks normal and also in the Console I can see the Orchestration Group Member State "Reboot required"

a few minutes later (about 15 Minutes left on the reboot timer) I see that the first node has ran its post script without rebooting. The second node already ran its pre script and started installing updates.
In the Console I see the first node (still with an active reboot timer) showing as idle while the second node shows "In Progress"

I've chosen my timings / timeouts deliberately with lots of time to spare
The maint Window both machines are in is 6 hours long
the Orchestration Group Timeout is 5 hours
each Member gets 2 hours
the pre and post scripts have a timeout of 1800 (I assume it's in seconds, as shown in the Orchestration Group Properties and not minutes as it is shown in the "Scripts" tab when selecting an orchestration group)

 

[angrygoat]

as a small update: The last answer from MS Support in our ticket was: "We're trying to reproduce it in our lab and will notify you".
This was 2 weeks ago and meanwhile I've sent the logs of 4 more failed orchestration procedures.