AIX 5L SysAdmin II: (Unit 05) – Disk Management Theory (Part 1 of 3)

Unit Objectives:
————————
1. Describe where LVM information is kept
2. Solve ODM-related LVM problems
3. Set up Mirroring
4. Explain the Quorum Mechanism
5. Describe Physcical Volume States

LVM Terms:
——————
PV = /dev/hdisk0
Regular VG = 32 PVs
Big VG = 128 PVs
PP = 32MB

Note: A Physical Partition size defines the smallest size of a FileSystem. Increasing the size of a file system then has to be in 32MB chunks.

(PP) Physical Partitions – (LP) Logical Partitions (Pointers to a PP)
(PV) Physical Volumes – (LV) Logical Volume
(VG) Volume Group

Note: A physical volume must all be assigned to one volume group.

cfgmgr – will find the device and give it a name

Logical Volumes:
1. Boot Logical Volume
2. JFS – Journaled File Systems
3. JFS Logs
4. Paging Space
5. Raw Devices

Volume Group Limits:
——————————–
Normal Volume Groups (mkvg)
#ofDisks Max.#ofPartitions/disk:
1 32512 mkvg -t 6
2 16256 mkvg -t 5
4 8128 mkvg -t 4
8 4064 mkvg -t 3
16 2032 mkvg -t 2
32 1016

Big Volume Groups (mkvg -B)
#ofDisks Max.#ofPartitions/disk:
1 130048
2 65024
4 32512
8 16256
16 8128
32 4064
64 2032
128 1016

mkvg -t 2
chvg -t

Mirroring: (Slower, duplicate copies of the data)
————–
Logical Volume (1 Logical Partition) –> 2 Physical Partitions

Striping: (Faster read/write access, if one drive fails can’t retrieve data)
————
Striped Logical Volume –> Stripe Units
hdisk0 (1,4,7)
hdisk1 (2,5,8)
hdisk2 (3,6,9)

Mirroring and Striping with RAID:
————————————————
RAID – Redundant Array of Independent Disks
RAID Adapter & Controller (Group of disks)

RAID Levels You Should Know About:
——————————————————-
Level – 1 (Striping)
Data is split into blocks. These blocks are written to or read from a series of disks in parallel. No data redundancy.

Level – 2 (Mirroring)
Data is split into blocks and duplicate copies are kept on separate disks. If an disk in the array fails, the mirrored data can be used.

Level – 5 (Striping with parity drives)
Data is split into blocks that are stripted across the disks. For each block parity information is written that allows the reconstruction in case of a disk failure.

FAQs:
———-
Q: Does AIX supports RAID level 5 at the software level?
A: No AIX doesn’t not support RAID level 5 at the software level.? If you wish to implement RAID 5 you must have the appropriate hardware.
?????
Q: Are the LVM identifiers unique worlwide?
A: Yes, the LVM identifiers are unique worldwide.? They are dependent on multiple factors such as machine ID which is always a unique ID number for each p-series server world-wide and the date and time the particular device or logical device is configured.

Q: Is all the information about the LVM stored in the ODM??
A: No, the information is stored in different object classes of the ODM, as well as, the VGDAs,? files through out the / file system such as /dev definitions, and Logical Volume Control Blocks.? This pretty much provides the ability to correct inconsistent information with the information in different areas of the system.

Q: What happens if a volume group doesn’t have quorum?
A: If a volume group doesn’t have quorum and is active AIX will close the volume group preventing access to any logical volumes in the affected volume group.? If the volume group is not active the varyonvg command will fail.? In either case, you can still force the varyon of the volume group with the varyonvg -f vgname command.? This action will force the varyon of the volume group and the disk or disks preventing the quorum will appear to be in the removed status.? This status can be confirmed by using the lspv hdisk# command.? Soon after that, test the devices by checking simple points of failure such as? powering on the PVs.? Sometimes lost of power is the reason for the loss of quorum and if that is the case then try once again varyonvg vgname.? If the command fails once again, then issue the command chpv -v a /dev/hdisk# to announce that the failure has been addressed and the PV or PVs should be active.? By issuing that command the next varyonvg command should work successfully if it was a temporary hardware failure.? If it doesn’t and you are still receiving hardware errors it may be necessary to replace the PV or PVs and those steps are listed in unit 6.

Q: What should I do if I noticed that there are inconsistencies with the ODM and the actual definitions of devices or logical devices on my system?
A: If you notice inconsistencies with the ODM and these are not addressable by running LVM commands then varyoff the volume group, export it and import it againg to clean up the inconsistencies in the ODM.? If it’s rootvg, you can not export the volume group.? Thus it will be necessary to execute the commands listed in the ‘rvgrecover’ script which was part of the discussion in unit 5.

Leave a Reply

Your email address will not be published. Required fields are marked *

*