HP 3PAR InForm OS 2.3.1 Extended MU5 Release Notes (QL226-96621, May 2013)

Table 1 Modifications to the InForm OS (continued)
DescriptionItemIssue ID
To avoid potential node panics during a Remote Copy re-synchronization operation,
the startrcopygroup -nosync command returns a failure in InForm OS 3.1.1
if an RC group is already started.
Remote Copy
re-synchronization
operation.
63602
Due to a compiler generating wrong assembly, a node is not able to take corrective
action when its message queue becomes full. This results in a failure of communication
Communication failure
between nodes results in
rolling upgrade revert.
63660
between the expedition and legacy nodes and a rolling upgrade is subsequently
reverted. This issue is resolved by adding a padding to the msg_q structure. To address
future upgrade failures faster, a debug crash dump on expedition nodes is taken if
there is a revert.
The firmware destroys its internal copy of scatter gather list, which results in the
firmware trying to access invalid or incorrect addresses. This is triggered only when
Core dumps with the
3PAR HBA
63759
the firmware uses DDR memory for BLP storage. If DDR memory is used at any time
in the past, there is a potential of the firmware polluting its internal memory (and thus
using invalid/incorrect host memory addresses) until the port is reset.
The underlying behavior that led to this potential instability is identified and corrected.On rare occasions a link
down event on a Remote
64109
Copy IP link can lead to
system unavailability.
The System Manager is updated to track the time taken to process SCSI-II and SCSI-3
reservation requests and to log any abnormally high processing times.
Record Timing
Information for
Reservations
64138
Issue the CAPACITY DATA HAS CHANGED' SCSI Unit Attention to hosts that use the
HP-UX host persona on VV grow. This allows recent versions of HP-UX to detect volume
resizing automatically.
Issue Capacity Change
Notification to HP-UX
hosts
64650
EnumerateInstances on TPD_PrivilegeForStorageHardwareID results in memory leak.EnumerateInstances on
TPD_PrivilegeForStorage
nl
65233
HardwareID results in
memory leak.
HidePaths sometimes fail with the error message "SPC does not exist" even though it
is there. This is resolved.
Resolved an issue where
the HidePaths operation
sometimes fails.
65299
When setting up a VV block, memory is allocated up front. If the VV block fails, this
memory may not be freed, which can cause a memory leak.
Memory leak if VV black
failure occurs
65460
Due to 0% charge level, the system cache is dropped to 0%. The battery that powers
RTC is identified to be root cause since it affects time (NULL) to return invalid values
for short period of time until it is corrected by network time server.
On rare occasions bad
RTC battery sets PR
failure timestamp way in
65463
future ultimately
degrading Battery
Backup Unit (BBU)
charge level to 0% and
preventing same from
improving.
There is an issue where each VV on a system ends up with the same VV master node
after a series of node down events. This results in the VV master node paging in a
Incorrect VV mastership
distribution after a series
of node down events.
65501
significant amount of meta data in order to restart the VVs and may result in the host
I/O timeout being exceeded.
In situations where there is a heavily loaded system with data-heavy snapshot with
heavily populated meta data tables, when deleting some of these snapshots, the meta
When deleting a large
and filled snapshot can
65646
data tables are traversed causing the CPU to be held for an excessive amount of time.lock up a CPU for a long
period of time. This can prevent other processes/tasks from running and can cause the background
internode comms processing to be delayed, resulting in the erroneous belief that one
of the other nodes has failed. This fix prevents the CPU from being held for an excessive
amount of time while traversing the meta data tables during snapshot removal.
8 HP 3PAR InForm Operating System Release Notes