Section 20.2.3, Excluding the NSS Cache Memory from Core Dumps
The encryption password is 6 to 16 standard ASCII characters. Make sure to employ security best practices for passwords. For information, see Section 31.15, Creating Strong Passwords.
Make sure to encrypt the data from an encrypted volume on backup media. Backups of an encrypted volume are not encrypted, unless it is a feature of the backup software you use. For information, see Section 31.4, Protecting Data During Backup and on Backup Media.
Make sure that you exclude the NSS cache memory from core dumps; otherwise, encrypted NSS volume data might be displayed in the clear. For information, see Section 31.5, Preventing Exposure of Sensitive Data in a Core Dump.
When working with encrypted volumes, it is important to realize that the volume password and key information is exchanged between user and kernel space as encrypted volumes are created and/or mounted. If you have logging enabled on the Linux server when you enter the encryption password, your password and volume key information might show up in the log file.
Even though the logging mechanisms are root user protected, we strongly recommend that you make sure logging is disabled when creating an encrypted volume or mounting the encrypted volume after a system reboot in order to protect the secrecy of your password credentials at these critical times when you are entering the encryption password.
For information, see Section 31.6, Preventing Exposure of the Encryption Password in a Log.
Direct I/O to an encrypted volume bypasses the EVS encryption engine and allows data to be stored in unencrypted format on the encrypted volume. This capability is useful for diagnostic, repair, or special-purpose applications, but should be avoided otherwise.
You should avoid using direct-I/O applications on encrypted volumes, especially for user data that you intend to be stored in encrypted format.
When you mount the shared volume and enter the password, NSS uses the password to create a key, which it stores in the server memory. The OES Cluster Services software passes the key to other nodes. After all servers hold the key, the volume is available while any one of the servers is still participating actively in the cluster. If all servers in the cluster fail, you must repeat this procedure when you recover the cluster and restart services.