AIX 6.1 allows you to mount a JFS2 filesystem without logging. This can increase performance by not requiring the FS to write metadata or log details. You could probably benefit by using this feature on temporary non-critical filesystems such as frequently used tmp space or a backup/restore tmp type of filesystem. With using this feature, you should assume that if the machine were to crash or you have other system related issues, you could potentially lose the entire non-logging filesystem.

If you still feel you can use this feature, you can mount by using the following:

To mount on the fly:

mount -o log=NULL /testfs

You will need to modify /etc/filesystems in order to make this change persistent across boots.

To demonstrate the simplicity of creating a Workload Partition within AIX 6.1 via CLI, here’s a quick example. The WPAR commands in AIX are pretty easy to figure out. You can even guess what they are, using the standard AIX file naming convention; mkwpar, lswpar, startwpar, stopwpar, chwpar, etc. If you prefer the GUI type of configuration, there is also a web based Workload Partition Manager tool that IBM introduced. Please see the AIX Version 6.1 IBM Workload Partitions Manager for AIX Redbook for more information.

Start out by creating the WPAR using the mkwpar command.

bash-3.00# mkwpar
mkwpar: 0960-046 Workload partition name is required.
Usage: mkwpar [-a] [-A] [-b devexportsFile] [-B wparBackupDevice] [-c]
[-d directory] [-D attr=value ...] … [-F] [-g vg ]
[-h hostName] [-l] [-L attr=value ...] [-M attr=value ...] …
[-N attr=value ...] … [-P] [-r] [-R attr=value ...]
[-S attr[+|-]=value …] [-s] [-u /path/to/script] [-v]
{ -n wparName [-p [name]] [-e existing_wpar | -f inFile] [-o outFile [-w]]
| -p name [-n wparName] [-e existing_wpar | -f inFile] [-o outFile [-w]]
| -f inFile [-n wparName] [-p [name]] [-o outFile [-w]]
| -w -o outFile [-n wparName] [-p [name]] [-f inFile | -e existing_wpar] }

Flags:
-a = Automatically resolve erroneous/conflicting settings.
-A = Start automatically on system boot.
-b = Path to permitted device exports file.
-B = Device or Pathname of savewpar backup image.
-c = Flag as checkpointable.
-d = Base directory.
-D = Override default device exports (devname, devtype, export).
-e = Get specification data from existing workload partition.
-f = Path to input specification file.
-F = Force - ignore certain errors.
-g = Default volume group for local file systems.
-h = Hostname.
-l = Create private, writeable versions of /usr and /opt.
-L = Logical volume management policy (image_data, shrink, ignore_maps).
-M = Mount settings (dev, directory, vfs, size, crfsopts, mode,
vg, logname, host, mountopts).
-n = Workload partition name.
-N = Network settings (interface, address, netmask, broadcast).
-o = Path to output specification file.
-p = Preserve file system data from the named mount group.
-P = Set workload partition root password interactively.
-r = Copy global network name resolution configuration into the workload
partition.
-R = Resource control settings (active, rset, CPU, memory, procVirtMem,
shares_CPU, shares_memory, totalProcesses, totalThreads).
-S = Configures the security settings of a workload partition (secfile, privs[+|-]).
-s = Start after creation.
-u = User script to execute on start & stop.
-v = Verbose mode.
-w = Only write specification file (do not create).

I am just creating a basic WPAR with a writable /usr and /opt. If -l is not specified, this WPAR will contain a loop back mounted fs for /usr and /opt using the global WPARs filesystem. You will obviously not be able to install any WPAR specific software that requires writing to /usr or /opt, unless your in the global WPAR. -a tells the WPAR to boot upon system boot, or global WPAR boot. -n will define the WPAR name, which shows up in the global WPAR. -h will define the hostname of the WPAR, and will grab its ethernet address from /etc/hosts.

bash-3.00# mkwpar -a -h wpar1 -l -n wpar1 -v
mkwpar: Creating file systems…
/
/home
/opt
/proc
/tmp
/usr
/var
Mounting all workload partition file systems.
mkwpar: Copying base files…
…
syncroot: Installing RPM file: /var/ssl/openssl.cnf
syncroot: RPM root packages are currently synchronized.
syncroot: Root part is currently synchronized.
syncroot: Returns Status = SUCCESS
Workload partition wpar1 created successfully.
mkwpar: 0960-390 To start the workload partition, execute the following as root: startwpar [-v] wpar1

mkwpar will create new logical volumes for the WPAR to live on. It will then make a copy of the running AIX instance and prepare the WPAR for boot.

’startwpar’ is the command you will boot the WPAR with.

bash-3.00# startwpar -v wpar1
Starting workload partition wpar1.
Mounting all workload partition file systems.
Mounting /wpars/wpar1
Mounting /wpars/wpar1/home
Mounting /wpars/wpar1/opt
Mounting /wpars/wpar1/proc
Mounting /wpars/wpar1/tmp
Mounting /wpars/wpar1/usr
Mounting /wpars/wpar1/var
Loading workload partition.
…
Exporting workload partition devices.
Starting workload partition subsystem cor_wpar1.
0513-059 The cor_wpar1 Subsystem has been started. Subsystem PID is 876724.
Verifying workload partition startup.
Return Status = SUCCESS.

Once the WPAR is booted, you can check the status by using ‘lswpar’

bash-3.00# lswpar
Name State Type Hostname Directory
—————————————————————-
wpar1 A S wpar1 /wpars/wpar1

Now that the WPAR is online, you can telnet to it, ssh to it, or use the ‘clogin’ tool for console login. You can also use clogin with a command as a parameter as well, then it will execute the command within the WPAR as shown.

bash-3.00# clogin wpar1
*******************************************************************************
* *
* *
* Welcome to AIX Version 6.1! *
* *
* *
* Please see the README file in /usr/lpp/bos for information pertinent to *
* this release of the AIX Operating System. *
* *
* *
*******************************************************************************

# hostname
wpar1
#

bash-3.00# clogin wpar1 hostname
wpar1

Tired of playing with it? ’stopwpar’ will kill it.

bash-3.00# stopwpar
Usage: stopwpar [-F|-h] [-N|-t ] [-r] [-v] wparName

Flags:
-F = Forced stop.
-h = Hard stop.
-N = No timeout for halt.
-t = User specified halt timeout in seconds.
-r = Reboot after halt is completed.
-v = Verbose mode.
bash-3.00# stopwpar wpar1
Stopping workload partition wpar1.
Stopping workload partition subsystem cor_wpar1.
0513-044 The cor_wpar1 Subsystem was requested to stop.
stopwpar: 0960-261 Waiting up to 600 seconds for workload partition to halt.
Shutting down all workload partition processes.
Unmounting all workload partition file systems.
bash-3.00#

Also you might want to check out ‘chwpar’ for changing WPAR configurations.

bash-3.00# chwpar
Usage:
To change a setting:

chwpar [-a] [-A] [-c] [-d directory] [-D attr=value ...] …
[-F] [-h hostName] [-n newName] [-N attr=value ...] …
[-P] [-R attr=value ...] [-S attr[+|-]=value …] [-u userScript] [-v]
wparName

To delete a setting:

chwpar -K [-A] [-c] [-D devname=value] … [-F] [-N address=value] …
[-R [attr ...] ] [-S] [-u] [-v]
wparName

Flags:
-a = Automatically resolve erroneous/conflicting settings.
-A = Start automatically on system boot.
-c = Flag as checkpointable.
-d = Base directory.
-D = Device exports (devname, devtype, export).
-F = Force - ignore certain errors.
-h = Hostname.
-n = New workload partition name.
-N = Network settings (interface, address, netmask, broadcast).
-P = Set workload partition root password interactively.
-R = Resource control settings (active, rset, CPU, memory, procVirtMem,
shares_CPU, shares_memory, totalProcesses, totalThreads).
-S = Configures the security settings of a workload partition (secfile, privs[+|-]).
-u = User script to execute on start & stop.
-v = Verbose mode.

That’s a basic overview of creating a WPAR. Consult the man page for more complex configurations. This just demonstrates basic functionality.

This is just a quick overview of LVM2 and how to use it. LVM in general makes managing servers much easier due to being more flexible with storage devices. Disk management is much simpler. No longer do you have to use an entire disk partition for a file system, you can just create a new logival volume and go on.

You can see the partition layout of this box. I’ll be playing around with /dev/hda6 which is a single partition labeled as Linux/LVM.

Device Boot Start End Blocks Id System
/dev/hda1 1 609 4891761 83 Linux
/dev/hda2 610 734 1004062+ 82 Linux swap / Solaris
/dev/hda3 735 1951 9775552+ 83 Linux
/dev/hda4 3168 19457 130849425 5 Extended
/dev/hda5 8034 19457 91763248+ 7 HPFS/NTFS
/dev/hda6 3168 4384 9775521 8e Linux LVM

The first step to using LVM is assigning an unused partition or physical disk to LVM. You do this by running the ‘pvcreate’ command. The device can be a disk partition, entire whole disk, meta device, or a loop back file.

firmo log # pvcreate /dev/hda6
Physical volume “/dev/hda6″ successfully created

firmo log # pvs
PV VG Fmt Attr PSize PFree
/dev/hda6 lvm2 — 9.32G 9.32G

Once the physical volume is initialized, you will have to create a volume group using the ‘vgcreate’ command. The volume group is a collection of physical volumes as created above. Your logical volumes will live within this volume group. It’s generally best to make a volume group for a certain group of storage since it will be somewhat sharing the same backend devices, such as rootvg in the AIX world. If you were to eventually create another volume group for a groupd of database filesystems, you would probably name it accordingly. One of the reasons I generally keep storage groups separate is for data migrations. If you have SAN storage going to host1, and have a single volume group with 10 different applications running in it, it’ll make migration of one or two applications very difficult when moving to a new host. If you have a volume group for each different application, you can easily manipulate that individual volume group to export it to a new host.

Here is an example of creating volume group test_vg using physical volume /dev/hda6

firmo ~ # vgcreate test_vg /dev/hda6
Volume group “test_vg” successfully created

The ‘vgs’ command will show you detected volume groups.

firmo ~ # vgs
VG #PV #LV #SN Attr VSize VFree
test_vg 1 0 0 wz–n- 9.32G 9.32G

Now that you have a usable volume group, you will need to create a logical volume to actually write the data to. This is done using the ‘lvcreate’ command as shown below.

This command is creating a 2GB logical volume that is read/write using test_vg to grab its storage from.

firmo linux # lvcreate -L 2G -p rw test_vg
Logical volume “lvol0″ created

Notice I didn’t specify a name, so the logical volume is named lvol0 by default. If you were to do this again, you could have lvol1, etc.

firmo linux # lvs
LV VG Attr LSize Origin Snap% Move Log Copy%
lvol0 test_vg -wi-a- 2.00G

Here I specify name for the LV.

firmo linux # lvcreate -L 2G -p rw -n test_lv test_vg
Logical volume “test_lv” created

firmo linux # lvs
LV VG Attr LSize Origin Snap% Move Log Copy%
lvol0 test_vg -wi-a- 2.00G
test_lv test_vg -wi-a- 2.00G

Now the device is ready for use. I will format it with reiserfs.

firmo linux # mkreiserfs /dev/test_vg/test_lv

Once the file system is created, you are free to mount the volume. Notice the mount device, it will be /dev/volume_group/logical_volume.

firmo linux # mount /dev/test_vg/test_lv /test
firmo linux # df -h /test
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/test_vg-test_lv
2.0G 33M 2.0G 2% /test

Once you’re done playing around, you can go ahead and delete the logical volume and volume group. You can even remove the physical volume definition for that matter.

The volume must be unmounted before removing the device as shown.

firmo linux # lvremove /dev/test_vg/test_lv
Can’t remove open logical volume “test_lv”
firmo linux # umount /test
firmo linux # lvremove /dev/test_vg/test_lv
Do you really want to remove active logical volume “test_lv”? [y/n]: y
Logical volume “test_lv” successfully removed

firmo linux # lvremove -f /dev/test_vg/lvol0
Logical volume “lvol0″ successfully removed

Now I remove the test_vg volume group

firmo linux # vgremove test_vg
Volume group “test_vg” successfully removed

And remove the physical volume

firmo linux # pvremove /dev/hda6
Labels on physical volume “/dev/hda6″ successfully wiped

Advanced POWER Virtualization on IBM System p Introduction and Configuration

Here’s a good link that I came across for using ssh-agent. There good .profile examples near the bottom.
http://mah.everybody.org/docs/ssh

Oddly enough, AIX 6.1 SP2 is now available. Sorta concerns me after seeing the patch list but hey, at least the bugs are being discovered and fixed. It’s kind of hard to imagine that 6.1 was just released last month, as well as SP1 during same time. Now we’re in December, a month later, and SP2 is out with a *lot* of fixes, hmmm. I guess I’m glad I currently have no plans for it anywhere except in the lab!

Check the link to Fix Central out for more information:

http://www-912.ibm.com/eserver/support/fixes/fixcentral/pfixpacks/61

So you have an LVM based file system that needs extended? Now a days this can be done pretty quickly with a few commands. The following will walk you through a basic online resize of an ext3 and reiserfs based filesystem. Keep in mind, if you’re dealing with critical production data of any type, you probably should resize those offline, and during a scheduled maintenance window as a safety measure.

#df -k /mnt/test
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/testvg-test
2097084 1760768 336316 84% /mnt/test

When resizing volumes, you have to keep in mind that the file system also has to be made aware of the new volume size. A common misconception is that you can resize the LVM device and the file system will automatically grow with it, not the case in ext3 and reiserfs that I’m aware of.

The test volume we’re messing with is nearly full, so we will add another 1GB to this. We’ll start off first by extending the logical volume using the ‘lvextend’ command.

#lvextend
Please specify either size or extents (not both)
lvextend: Add space to a logical volume

lvextend
[-A|--autobackup y|n]
[--alloc AllocationPolicy]
[-d|--debug]
[-h|--help]
[-i|--stripes Stripes [-I|--stripesize StripeSize]]
{-l|–extents [+]LogicalExtentsNumber[%{VG|FREE}] |
-L|–size [+]LogicalVolumeSize[kKmMgGtTpPeE]}
[-m|--mirrors Mirrors]
[-n|--nofsck]
[-r|--resizefs]
[-t|--test]
[--type VolumeType]
[-v|--verbose]
[--version]
LogicalVolume[Path] [ PhysicalVolumePath... ]

#lvextend -L+1G /dev/testvg/test
Extending logical volume test to 3.00 GB
Logical volume test successfully resized
#lvs /dev/testvg/test
LV VG Attr LSize Origin Snap% Move Log Copy%
test testvg -wi-ao 3.00G
#
#df -h /mnt/test
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/testvg-test
2.0G 1.7G 329M 84% /mnt/test
#

Note the file system size remains at 2GB. Once the logical volume is extended, we will have to figure out what type of file system this is. You can run the following to determine this:

#mount | grep /mnt/test
/dev/mapper/testvg-test on /mnt/test type reiserfs (rw)

So we’re dealing with reiserfs.

It may also be smart to run an fsck at this point, depending on whether or not you can take the volume offline. Since this is test data, I’ll ignore the fsck and go ahead and resize the file system while it’s online. For reiserfs, this is done using the ‘resize_reiserfs’ command. You can either specify the size of let the command figure it out, and grow to the full extent of the volume. I’ll let it figure it out.

#resize_reiserfs /dev/testvg/test
resize_reiserfs 3.6.19 (2003 www.namesys.com)

resize_reiserfs: On-line resizing finished successfully.

#df -h /mnt/test/
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/testvg-test
3.0G 1.7G 1.4G 56% /mnt/test
#

That’s it… Pretty tough huh?

Now for ext3, we’ll use the same volume.

#df -h /mnt/test
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/testvg-test
2.0G 68M 1.9G 4% /mnt/test
#mount | grep /mnt/test
/dev/mapper/testvg-test on /mnt/test type ext3 (rw)
#

#lvextend -L+1G /dev/testvg/test
Extending logical volume test to 3.00 GB
Logical volume test successfully resized
#!df
df -h /mnt/test
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/testvg-test
2.0G 68M 1.9G 4% /mnt/test
#

You’ll resize ext3 using the ‘resize2fs’ command as shown. Pretty similar to reiserfs.

#resize2fs
resize2fs 1.40.2 (12-Jul-2007)
Usage: resize2fs [-d debug_flags] [-f] [-F] [-p] device [new_size]

#resize2fs /dev/testvg/test
resize2fs 1.40.2 (12-Jul-2007)
Filesystem at /dev/testvg/test is mounted on /mnt/test; on-line resizing required
old desc_blocks = 1, new_desc_blocks = 1
Performing an on-line resize of /dev/testvg/test to 786432 (4k) blocks.
The filesystem on /dev/testvg/test is now 786432 blocks long.

#!df
df -h /mnt/test
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/testvg-test
3.0G 68M 2.8G 3% /mnt/test
#

A good redbook on Live Partition Mobility

IBM System p Live Partition Mobility

Some good information in this redbook if you work with SVC at all.

SAN Volume Controller: Best Practices and Performance Guidelines