On IBM P5 505 (maint level 5300-07) servers there is a concurently accessed
external storage, which is connected using external Ultra320 SCSI port (DAS).
External storage is detected as hdisk2 which belongs to volume group dbvg.
Volume group dbvg is used as a storage for Oracle RAC 10gR2 solution.
Here is information about volume group dbvg
:
[admin@node1 ~]$ lsvg dbvg
VOLUME GROUP: dbvg VG IDENTIFIER: 0004523a0000d3000..
VG STATE: active PP SIZE: 128 megabyte(s)
VG PERMISSION: read/write TOTAL PPs: 3725 (476800 mega..
MAX LVs: 256 FREE PPs: 0 (0 megabytes)
LVs: 111 USED PPs: 3725 (476800 mega..
OPEN LVs: 64 QUORUM: 1 (Disabled)
TOTAL PVs: 1 VG DESCRIPTORS: 2
STALE PVs: 0 STALE PPs: 0
ACTIVE PVs: 1 AUTO ON: no
Concurrent: Enhanced-Capable Auto-Concurrent: Disabled
VG Mode: Non-Concurrent
MAX PPs per VG: 32768 MAX PVs: 1024
LTG size (Dynamic): 256 kilobyte(s) AUTO SYNC: no
HOT SPARE: no BB POLICY: relocatable
[admin@node1 ~]$
If to list logical volumes (which is 111, so I will be showing only few here) we would see all of those residing on one physical volume:
[admin@node1 ~]$ lsvg -l dbvg
dbvg:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
...
sysaux jfs2 8 8 1 open/syncd N/A
system jfs2 8 8 1 open/syncd N/A
ocr1 jfs2 2 2 1 open/syncd N/A
ocr2 jfs2 2 2 1 open/syncd N/A
vote1 jfs2 1 1 1 open/syncd N/A
vote2 jfs2 1 1 1 open/syncd N/A
vote3 jfs2 1 1 1 open/syncd N/A
sub_1 jfs2 41 41 1 open/syncd N/A
etc_1 jfs2 41 41 1 open/syncd N/A
...
[admin@node1 ~]$
Currently working connection diagram is as below:
|---------------------------------DIAGRAM #1-----------------------------------|
|----------------Currently used external storage connection diagram------------|
┌───────┬────────────┐ ┌───────┬────────────┐
│#node1 │ IBM P5 505 │ │#node2 │ IBM P5 505 │
├───────┴────────────┤ ├───────┴────────────┤
│ VG rootvg │ │ VG rootvg │
│ │ │ │
│ VG dbvg │ │ VG dbvg │
│ │ ┌────────────────┤ ├────────────────┐ │ │
│ └─┤ PV hdisk2 │ │ PV hdisk2 ├─┘ │
│ │ ultra320 SCSI │<───(scsi)──────┐ ┌──────(scsi)───>│ ultra320 SCSI │ │
│ └────────────────┤ │ │ ├────────────────┘ │
└────────────────────┘ │ │ └────────────────────┘
│ │
┌─────────────────────────────┐ │ │
│ SCSI-to-SATA │ in ch A │<──────┘ │
│ JBOD enclosure ├──────────┤ │
│ │ in ch B │<─────────┘
│ Single RAID ├──────────┤
│ controller │ out ch A │<──(terminator)
│ ├──────────┤
│ │ out ch B │<──(terminator)
└──────────────────┴──────────┘
|-----------------------------END OF-DIAGRAM #1--------------------------------|
However the problem is that external storage is vulnarable due to it has single RAID controller and upon it's failure external storage would become inaccessible even hard disks will be ok inside.
To solve this, there is a plan to add additional physical volume (by using iSCSI adapter) to the volume group dbvg
and mirror logical volumes over two physical volumes. This should give something like:
[admin@node1 ~]$ lsvg -p dbvg
dbvg:
PV_NAME PV STATE TOTAL PPs FREE PPs FREE DISTRIBUTION
hdisk2 active 3725 0 00..00..00..00..00
hdisk3 active 3725 0 00..00..00..00..00
[admin@node1 ~]$
And after executing mklvcopy <LV_name> <copy_number> <destination__PV>
for all logical volumes we should see something like:
[admin@node1 ~]$ lsvg -l dbvg
dbvg:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
...
sysaux jfs2 8 16 2 open/syncd N/A
system jfs2 8 16 2 open/syncd N/A
ocr1 jfs2 2 4 2 open/syncd N/A
ocr2 jfs2 2 4 2 open/syncd N/A
vote1 jfs2 1 2 2 open/syncd N/A
vote2 jfs2 1 2 2 open/syncd N/A
vote3 jfs2 1 2 2 open/syncd N/A
sub_1 jfs2 41 82 2 open/syncd N/A
etc_1 jfs2 41 82 2 open/syncd N/A
...
[admin@node1 ~]$
In such situation dbvg volume group would be mirrored accross SCSI and iSCSI physical volumes and thus failure-proof situation would be achieved. Planned diagram would be then:
|---------------------------------DIAGRAM #2-----------------------------------|
|----------------Planned external storage mirror on SCSI + iSCSI---------------|
┌───────┬────────────┐ ┌───────┬────────────┐
│#node1 │ IBM P5 505 │ │#node2 │ IBM P5 505 │
├───────┴────────────┤ ├───────┴────────────┤
│ VG rootvg │ │ VG rootvg │
│ │ │ │
│ VG dbvg │ │ VG dbvg │
│ │ ┌────────────────┤ ├────────────────┐ │ │
│ ├─┤ PV hdisk2 │ │ PV hdisk2 ├─┤ │
│ │ │ ultra320 SCSI │<───(scsi)────┐ ┌────────(scsi)───>│ ultra320 SCSI│ │ │
│ │ └────────────────┤ │ │ ├────────────────┘ │ │
│ │ ┌────────────────┤ │ │ ├────────────────┐ │ │
│ └─┤ PV hdisk3 │ │ │ │ PV hdisk3 ├─┘ │
│ │ iSCSI adapter │<───(eth)────────────┐ ┌──(eth)───>│ iSCSI adapter │ │
│ └────────────────┤ │ │ │ │ ├────────────────┘ │
└────────────────────┘ │ │ │ │ └────────────────────┘
│ │ │ │
┌─────────────────────────────┐ │ │ │ │
│ SCSI-to-SATA │ in ch A │<────┘ │ │ │
│ JBOD enclosure ├──────────┤ │ │ │
│ │ in ch B │<───────┘ │ │
│ Single RAID ├──────────┤ │ │
│ controller │ out ch A │<──(term) │ │
│ ├──────────┤ │ │
│ │ out ch B │<──(term) │ │
└──────────────────┴──────────┘ │ │
│ │
┌─────────────────────────────┐ │ │
│ SAN / iSCSI storage │<───────────┘ │
│ │ │
│ │<──────────────┘
└─────────────────────────────┘
|-----------------------------END OF-DIAGRAM #2--------------------------------|
So the question is - is it ok from general AIX OS and IBM hardware perspective to have SCSI and iSCSI physical volumes inside same volume group and have mirroring on those?