You should be aware of changes in Data ONTAP behavior that might occur if you upgrade to Data ONTAP 8.2 or later releases.
Be sure to check the
Known Problems and Limitations section in the Release Notes for your target Data ONTAP release to see a complete list of changes in behavior after upgrade to the target release.
This list summarizes significant changes known at publication time:
- Storage system disruption occurs when there are no asynchronous messages to dispatch tasks between different sub-systems within Data ONTAP.
The number of asynchronous messages per storage system is limited.
- If you are configuring volumes with zero fractional reserve and using certain technologies or Data ONTAP features, you must take some extra precautions to avoid out of space errors.
- The
cf_remote
license
is installed by default on all new
32xx
and
62xx
storage systems for
versions of Data ONTAP earlier than 8.2.
If you upgraded to Data
ONTAP 8.2 and are using mirrored aggregates that are not in a
MetroCluster configuration, the presence of this license can reduce
read performance because data is read from only one plex.
- If
V-Series system
FC initiator ports are zoned with
Hitachi or HP XP array target ports before the storage array
parameters are set and the LUNs are mapped to the host groups, you
might not be able to see any LUNs presented to the Data ONTAP
interface.
- In a single path SnapMirror transfer, you cannot
use the
interface.blocked.mgmt_data_traffic
option to block SnapMirror traffic from e0M on the
destination storage system.
You can use this option to block
SnapMirror transfer only on the source storage system.
- When an interface name used for configuring the
partner interface is applied as a host name alias in the /etc/hosts file and the interface name is used in an ifconfig command, the host name gets resolved in the local host
context and not the partner context, resulting in unavailability of
the partner node during takeover.
- Fast path is enabled by default on all storage systems.
However, you might have to disable fast path in certain scenarios to avoid issues such as performance degradation or failed software upgrades.
- If a CIFS client with a NetBIOS name of more than 16 characters attempts to connect to the storage system, authentication fails.
- If domain controllers become unavailable to a vFiler unit, pending CIFS authentication requests accumulate on the storage system.
If the issue is not resolved, this can prevent other vFiler units on the storage system from successfully completing CIFS authentication requests.
- On a newly installed storage controller without CIFS configured, if you create a qtree on a volume that has security style set to NTFS, the resulting qtree has NTFS as the security style and UNIX as the effective style.
This means that UNIX clients can set attributes on the NTFS volumes, which normally would not be allowed.
- When you create
SnapVault relationships for SnapLock secondary
volumes
in vfiler0, you also manually set
the
snapvault.lockvault_log_volume
option to the name of a
SnapLock for SnapVault
log
volume.
After this option is set, SnapVault cannot be enabled on
any vFiler units on the storage system.
- After upgrading to Data ONTAP 8.2.0 or later releases, some Oracle Linux-based systems using ASMlib can fail to recognize disk groups.