You can not decrypt the key.
https://www.ciscolive.com/c/dam/r/ciscolive/emea/docs/2024/pdf/BRKEWN-2339.pdf
> is there a way to retrieve it?
Not that I have heard of but I suppose it would necessarily mean attaching some sort of analyser directly to motherboard or extracting the NVRAM as has been demonstrated on some PC exploits.
> Are you implying that the encryption key itself is encrypted when stored in the nvram?
We don't know. For obvious reasons all Cisco says is that it is stored "securely" in NVRAM. But like Flavio I expect it would have some sort of encryption at the very least, which would have to be reversible, but known only to Cisco developers who wrote the code. It would probably use something fixed to the device (like a serial number or similar but not easy to find like serial number) as a salt.
> I've tried "no service password-encryption" and "no password encryption aes", but when doing a show run, the passwords are still Type 8 encrypted....
AES encrypted keys and passwords are type 6. Old Cisco reversible password encryption is type 7.
Types 4, 5, 8 & 9 are hashes not encryptions. They can be cracked but not decrypted.
4 was proven to be fundamentally insecure, 5 uses MD5 now considered easily crackable, 8 & 9 are the stronger variants of 4 and 5. Type 8 is hashed using PBKDF2, SHA-256, 80-bit salt, 20,000 iterations. Type 9 uses the SCRYPT hashing algorithm (RFC 7914) 80-bit salt, 16384 iterations. There's some debate but type 9 is mostly considered best practice for hashes on Cisco IOS although as per below NSA currently recommends type 8.
https://community.cisco.com/t5/networking-knowledge-base/understanding-the-differences-between-the-cisco-password-secret/ta-p/3163238 has more detail.
From https://media.defense.gov/2022/Feb/17/2002940795/-1/-1/1/CSI_CISCO_PASSWORD_TYPES_BEST_PRACTICES_20220217.PDF
So even if you knew the AES key that would not help with the type 8 passwords because they are not encrypted with that.
So your only hope for those is a PC with plenty of CPU/GPU and running them against one of the freely available cracking tools (HashCat) to see what possible values come up and you might instantly recognise one of the results.
> 9800-Lab(config)#key config-key password encrypt key testing
> 9800-Lab(config)#key config-key password encrypt key testing123
> Old key: <<I input "testing" here
> %% Key length less than 8 chars
<facepalm> That is just classically incompetent and sloppy coding but I guess we're getting used to seeing that kind of thing - at least now you know minimum length 8 characters <smile>
> Also, if I ever can get the encryption keys to match on each WLC, are the encrypted passwords shown in the running configs supposed to match between different WLCs?
I don't think they necessarily look identical when generated on different boxes but you can copy them from one box to another as long as you have used the same master AES key on every box before applying them. Previously you could apply them before configuring the master key and they would start working as soon as the correct master key was configured (right now VPN keys still work this way) but Cisco have introduced logic (intended to be applied to VPN keys too in near future) that it won't accept an encrypted key unless it can be decrypted with the currently configured AES master key which is why you must configure the AES key before applying the encrypted passwords.
