I wanted to do some DR Testing on a UCS environment I'm shortly placing into production. The system gets backed up daily using UCS Powertools.
To start the DR test I simply erased the configuration on BOTH Fabric Interconnects - running f/w 2.2(5b) by the way. Each FI was then configured with the same IP addresses etc and the cluster was eventually brought up and I could log into UCSM. Note the admin password differs this time around.
I then proceeded to immediately try and restore the config-all.xml file that had been created, but this failed with the error "Encrypted password length incorrect (possibly corrupted)".
I AM able to restore a config-all.xml file that has been subsequently created.
Technically I should be able to do what I'm doing, right? I have the full state backup, too, but that would involve a bit more messing about and I understood it just to be another option in the recovery arsenal.
Has anyone done what I'm doing and had it work?
The URL posted the bug exactly what you encountered and provided the workaround.
Since I didn't want to modify my xml config, so my workaround was to downgrade the UCSM to older version 2.2(3c). It was then successfully to import my xml file without any issue. After that, I activated the UCSM version back to 2.2(5b).
Awesome, thanks very much for that! I never considered it could be a bug.
The bug report page does say it's fixed in 2.2(6c) but I cannot find any mention of it in the release notes for this version so, given this particular UCS instance is pre-production I'll upgrade and report back.
2.2(6d) works with regard to backing up and restoring XML by the way. TAC verified it was the bug Mac Chung Lok posted.