Considerations for stateless protocols

Configurations that include client connections using stateless NAS and SAN protocols generally do not experience adverse effects during upgrades if the clients are configured according to recommended guidelines.

If you are using stateless protocols, consider the following:

• NFS hard mounts

  No adverse behavior is experienced on the clients during upgrade. Clients might receive some messages similar to the following until the node reboots: NFS server not responding, retrying

  In general, read/write directories should be hard-mounted. Hard mounts are the default type of mount.

• NFS soft mounts

  You should not use soft mounts when there is a possibility of frequent NFS timeouts. Race conditions can occur as a result of these timeouts, which can lead to data corruption. Furthermore, some applications cannot properly handle errors that occur when an NFS operation reaches a timeout using soft mounts.

  Situations that can cause frequent timeouts include nondisruptive upgrades or any takeover or giveback event in an HA configuration.

  In general, soft mounts should be used only when reading solely from a disk; even then, understand that any soft mount is unreliable.

• SAN protocols

  No adverse behavior is experienced on FC or iSCSI clients if they are configured according to the recommended guidelines listed in the Lenovo Storage Interoperation Center (LSIC).