HP-UX Logical Volume Manager and MirrorDisk/UX Release Notes HP-UX 11i v3 (March 2011)

Table 1 LVM Fixes in HP-UX 11i v3 March 2011 Release (continued)
DescriptionDefect ID
In a configuration where large numbers of I/O are in progress for many logical volumes and
automatic increase of pre-allocated extents for a space-efficient snapshot is enabled, the
following message is logged in syslog:
lvmpud->lvextend[25174]: "LogicalExtentsNumber" is not bigger than
current setting.
vmunix: LVM: ERROR: The task posted for increasing the pre-allocated
extents failed for the snapshot LV (VG 128 0x004000, LV Number: 4
8).
QXCR1001039721
1
When creating the first snapshot for a logical volume while raw I/Os are in progress, the
system panics.
QXCR1001040067
1
The default threshold value for a striped space-efficient snapshot is reached when the number
of free extents is less than or equal to 1 full extent worth of stripes. In the case where the stripe
width (as specified by lvcreate -i) is two or small, the default threshold value is reached
when there are only 2 free pre-allocated extents. When this threshold is reached and if auto
pre-allocation is enabled, LVM tries to increase the pre-allocated pool size by only 2 extents,
which is very small in size.
If the I/O rate is faster than the pre-allocation, subsequent I/Os may make the snapshot as
over-committed and inoperative.
QXCR1001042095
1
When the LVM API lvml_session_init(), in the shared library liblvml.so, is invoked
from a threaded application, it fails with a non-zero return value.
QXCR1001043020
1
When attempting to halt an HP Serviceguard cluster, hosts panic with instruction TLB Fault.
During MORE operations. entries are added and deleted from the sndq linked list. Under
heavy use of MORE, it is possible for the first entry in the list to be deleted. When the cluster
is being shut down the queue is freed starting with the first entry, which has already been
freed.
QXCR1001044941
1
The vgcfgbackup command fails while extending a volume group with a cluster device
special file (cDSF) having a persistent DSF.
QXCR1001045693
1
When the server node is running nomwcsyncd, the system panics with the following message:
"slvmp: multiple block IO requests"
QXCR1001045832
1
The definition of 'Act PV' in the vgdisplay(1M) manpage is misleading.QXCR1001046984
A pvcreate command on the un-partitioned device special file (DSF) incorrectly succeeds
for a disk that has a partitioned DSF already in the lvmtab file.
QXCR1001047819
1
When pvdisplay is invoked for the cluster device special file (DSF) (/dev/cdisk/diskxx),
the Data_Start, Data_End and Bootable attributes are displayed incorrectly as
‘unavailable’.
QXCR1001047822
The vgscan -N command fails when legacy device special file (DSF) mode is disabled.QXCR1001047826
1
Sometimes lvdisplay incorrectly displays the status of a snapshot logical volume as stale
even though there are no stale extents in the snapshot logical volume.
QXCR1001047828
1
When the volume group metadata (which is not activated in shared mode) is updated by
different nodes, the system panics with the following stack trace:
IP Function Name
0xe0000000023322c0 bad_kern_reference+0xa0
0xe0000000008286d0 $cold_pfault+0x7e0
0xe000000000fd9e90 vm_hndlr+0x1590
0xe000000001f5a780 bubbledown+0x0
0xe0000001601bcf80 lvmp_vgsa_tst_pxstale+0xa0
0xe000000160098740 lvmp_init_lvols+0x820
0xe00000016009d0c0 lvmp_activatevg_common+0x480
QXCR1001049306
Known problems fixed in this version 5