Replace a failed Sup720?

[mdeloach]

Earlier this week we had a Sup720 fail. It was in a master/slave (SSO?) config and the backup properly took over running the switch.

Unfortunately the failed blade did not fail completely, and kept rebooting attempting to reclaim control and then rebooting again. Annoying.

We pulled the failed blade and received a RMA replacement, confirmed the model, swapped the flash card with the failed unit and attempted to install it.

It booted without major errors and seemed to switch into master mode, but in the process knocked all ports offline and out 10G blade sis not recover.

I've attempted to review the documentation to see what step I missed, as I clearly missed something, but I am not finding it.

How can I replace the blade without knocking all ports offline? Do I need to break the SSO arrangement and re-establish it after the blade boots?

[#2]

SRM with SSO allows Online insertion and removal (OIR) of the redundant Supervisor Engine 720 for maintenance. When the redundant Supervisor Engine 720 is inserted, the active Supervisor Engine 720 detects its presence and begins to transition the redundant Supervisor Engine 720 to a fully initialized state. If the image on the redundant Supervisor Engine 720 does not match the image on the active Supervisor Engine 720, RPR redundancy mode is used.

[mdeloach]

That is exactly what I would have expected, but it didn't work that way.

We are booting off the CF cards in the Disk0: slot. I removed the Disk0: CF card from the

failed unit and installed it in the new unit.

Upon inserting the replacement, every blade in the switch 'rebooted' and the 10G blade did not recover.

It's been a week since the primary failed and we have run without a back due to the hiccup during the last install

attempt. I will be trying again this Friday after work, so the interruption will be minimal if it does the same thing again.