MU3 Release Notes

ManualsBrandsHP ManualsComputer AccessoriesHP 3PAR Inform S800/4x400GB Magazine LTU

DescriptionItemBug ID

• Resolved table inconsistencies due to race condition between

power down and error handling

• Resolved an unexpected termination issue when there were

multiple link resets without power cycle

• Prevent microcode download of Toshiba flash base firmware

• Resolved the security issue so that the drive will not be

unlocked after the number of incorrect password attempts

exceed the limit.

• Improved the performance, time-wise, to complete a full

Windows format under disk management

FW 2806 contains the following changes:

• Resolved an issue that will cause background cleanup

operations to be skipped and therefore reduced the remaining

available blocks.

It is a standard policy to test BBU on all platforms every two

weeks to confirm that it is working fine. The test is automatically

Biweekly automatic Backup Battery Unit

(BBU) test by System Manager turns on

but never turns off

70712

conducted by System Manager by turning on BBU for 55 seconds.

On HP 3PAR P10000 platform, NEMOE conducts the same test

on behalf of System Manager by turning it on and off. But due

to a bug in NEMOE FW, the test is never turned off and BBU is

discharged indefinitely which leaves it irrecoverable and

permanently damaged. Also, if this happens on enough number

of nodes in the system, the cache will be dropped accordingly

and in worst case, the system can enter write-through mode

causing severe performance degradation. The enhancement in

NEMOE FW (3.47) prevents BBU from discharging if minimal

condition - at least one PS has AC ok - is met.

The driver issued a PLOGI while there is already a pending

PLOGI. This results in unstable DB state which in turn causes

Fibre Channel driver issue70828

LOGOs to go out to the remote side. To resolve this, terminate

the previously pending PLOGI before issuing a new PLOGI.

Certain internode messages need to allocate physically

contiguous large-chunk memory and the allocation may fail on

Better memory management to address

memory allocation failure

70906

a system with highly fragmented memory. Software now

addresses this issue by pre-allocating memory for those messages.

In earlier InForm OS versions, modification of an initiator's SCSI3

Persistent Reservation key is not handled correctly while that

Modification of Persistent Reservation

Key

71104

initiator held a reservation. The initiator's old reservation key is

reflected in the data returned in response to a PERSISTENT

RESERVE IN/READ REGISTRATION SCSI command and by the

output of the showrsv CLI command. The issue is addressed now

and the reported data remains consistent after modification of

the reservation key.

The initial default configuration of a CNA used to be ‘CNA’ type

in ‘suspended’ mode in a ‘config_wait’ state. The default

Initial default Converge Network

Adaptor (CNA) configuration set to

iSCSI target offline

71174

configuration is now changed to a ‘iSCSI’ type in ‘target’ mode

in an ‘offline’ state. This is done to avoid problems in running

checkhealth and admithw operations. This does not impact current

CNA configurations, only the starting point for newly installed

systems or newly added CNAs.

If virtual volume master node goes down while it is processing

multiple log pages for the same virtual volume family, the

Incorrect node-down recovery of

multiple processing log pages causes

double free

71857

node-down recovery should recover those log pages in one

bundle, but the new virtual volume master node may try to recover

those log pages in 2 bundles which cause double-free when it

tries to handle the second bundle of recovered log pages.

40 InForm OS 3.1.1 MU2 Release Notes