HP 3PAR OS 3.1.2 Release Notes

DescriptionItemIssue ID
Using CLI command controlport rst -l on a 10G iSCSI port
under heavy host I/O can lead to fatal PCI errors and a subsequent
node reboot.
CLI command controlport rst
-l can lead to fatal PCI errors
68387
Need to remove references to the MENU Driven Application from the
UG if it is deprecated.
User guide edits68839
Need to update RMO 4.1 RN's to indicate the Menu Driven
Application has been deprecated
Menu Driven Application
deprecated
68840
There is a problem with the chunklet moving algorithm used in the
intra-node (tunenodech) phase of tunesys. This causes a high
tunesys should not be run with a
maxchunk value of more than 2
(default)
70792
chunklet move failure rate if the value of maxchunk (number of
chunklets that can be moved in parallel) is set to greater than 2
(default). It is recommended that the value of maxchunk is left at its
default for all tunesys runs.
In this guide the maximum supported RCIP latency (delay times) is
incorrect.
Remote Copy User Guide71952
On FC ports with a large number of attached hosts, failover/failback
may not be sufficiently fast with the consequence that multipathing
Persistent Port fc failover/failback
SLES host device-mapper detected
fail paths
75801
messages may be seen in host logs. Host multipathing may temporarily
mark paths as failed but they are quickly re-instated.
This behavior has been observed on Linux hosts (Redhat and SLES11)
which have more stringent path validation algorithms. However, it is
conceivable that it could also be observed on other host types
depending on how heavily the interface is loaded.
Cisco VIC HBA M81KR is not supported with 3.1.2 at this time.I/O fails when a persistent port
event occurs with windows 2008
r2 sp1 on ucs with m81kr
79411
This problem is a defect on the Solaris host. The Brocade switch has
a feature where it will consolidate events that happen in close
Solaris host loses paths in some
failover/failback scenarios, in
80283
succession into a single RSCN. The FC subsystem on Solaris does notconfigurations involving a Brocade
switch handle this appropriately and therefore does not re-establish all the
paths as it should.
Impact: Assuming host multi-pathing is configured properly, if this
problem occurs the customer may lose some IO paths, resulting in
possible reduced performance until the failed paths are restored.
Failed paths can be restored via controlport rst.
Note that this problem only shows up with the combination of Brocade
switch and a Solaris 11 host.
A VV which is split between tiers by Adaptive Optimization may
incorrectly display RAID type and device type (SSD, FC, NL)
information for a secondary tier CPG.
A VV in an AO configuration may
occasionally display incorrect RAID
type and device type information
80428
This is a display issue and does not affect the primary tier CPG from
which new VV growth is taken.
Active volume removals will prevent periodic syncs starting for volumes
within the same volume family. An error message will be returned
Error message that volume is locked80779
indicating that the volume is locked. Once the removals have finished
the next periodic sync will succeed.
During an online upgrade periodic remote copy groups may stop
syncing. If this occurs, normal operation will resume as soon as the
Periodic RC stalled during online
upgrade from 3.1.1.MU1 to 3.1.2
81188
upgrade is completed (i.e. after the cli upgradesys –finish
command has been entered). There is no need to stop periodic groups
before the upgrade is carried out. Synchronous Remote Copy groups
are unaffected. I/O to volumes in remote copy groups is unaffected.
No manual intervention is required to Remote Copy groups or volumes
before, during or after the upgrade.
16 HP 3PAR OS 3.1.2 GA Release Notes