1 Reply Latest reply: Feb 13, 2018 6:49 AM by Ken Vance RSS

Steelhead CX upgrade to 9.2 and SN: S28250

Prashant Rojesara

Can I someone help to clarify deployment options to prevent QoS misclassification issues after an upgrade to 9.2.x as per solution document S28250?


At present, we have upgraded the SCC to 9.2.2. We have ~180 Steelheads currently on 9.1.3a to be upgrade to 9.2.3.


The configuration of Path Selection and QoS for all steelheads:

Path Selection - DISABLED

Enable Outbound QoS Shaping - DISABLED

Enable Inbound QoS Shaping - DISABLED

Enable Outbound QoS Marking - ENABLED


Manage QoS Per Interface:

Outbound and Inbound QoS - ENABLED


We are trying to upgrade the steelheads in a phased approach over a few weeks rather than big bang. the Head end units will be completed first.


The test lab has been upgraded to 9.2.3 by upgrading, disabling qos, rebooting, scc push then qos re-enable.


The solution document advises that "After ALL devices have rebooted to RiOS 9.2, perform a regular policy push to re-enable your existing QoS and Path Selection feature and rules settings.". Does this mean we have to have QoS disabled for the entire deployment timeline before re-enabling QoS as it mentions when ALL devices are on 9.2? Or can we just disable QoS before upgrade per steelhead and push policy then enable QoS on them individually or in groups as we work through the phased deployment?

Look forward to see response asap.

  • Re: Steelhead CX upgrade to 9.2 and SN: S28250
    Ken Vance



    its important to disable Qos and path selection before the upgrade (if I remember our upgrade to 9.2, it's just a couple of cli commands if I recall correctly).


    Upgrade the device then push policies to restore the Qos and/or path selection Configuration.


    A note of warning - in the lab we omitted the commands to disable Qos on a couple of steelheads before upgrading. We then had to delete the resulting corrupted config and manually copy/paste the correct updated Qos config from a successfully upgraded device. Not recommended!


    Personally,I took it to mean a batch at a time as we have too many to do in one hit (take this advice at your own risk)


    Get your core devices successfully upgraded, and their policies pushed,  then the batch of posts upgraded and push the policies to that batch.