[AIX] Increase Page Size on hd6 of RootVG

Today our AIX box just killed all the SSH connections and no one could log back into the machine. We could, howver, ping the machine and an nmap scan showed all necessary services were up, but they were just not responding.

After anylisys of the logs it was determined that:

1) There was a core dump produced
2) The firmware needed to be date from:  EL320_76 to: EL340_75
3) The paging size on hd6 of rootvg was not large enough 512M < 10G

I will address the last point (3) in this post. The  paging size on the system was set to 512M on a system that has 32GB real memory.

# lsps -a
Page Space      Physical Volume   Volume Group Size %Used Active  Auto  Type Chksum

hd6             hdisk0            rootvg         512MB     6   yes   yes    lv     0

Apparently all services stopped responding when paging started. When the machine came back online it was recommend that the paging size be increase to 10G. Since we already had a paging size of 512M, we just needed to increase by 9.5G.  But, in order to increase the paging size you need to check the PP size and the amout of free PPs. In my case I needed to increase by 160PPs. The PP SIZE is: 128 megabytes. Just convert 9.5G to MB and divide by the PP SIZE to determine how many PPs will give you 9.5GB. I needed 76 PPs.

Use smitty to increase the page size:

# smitty chps

Cheers,
-swinful

[Drivers] HP Office Jet 6310 All-In-One

I could not for the life of me locate online the HP drivers for my Office Jet 6310. Not even on the HP website! I checked both for PC and Mac drivers, but no luck. I checked the HP website below and even Google:

http://h10025.www1.hp.com/ewfrf/wc/softwareCategory?product=1119598&lc=en&cc=us&dlc=en&lang=en&cc=us

I even checked torrent sites with no luck! But, luckyly I was able to download the drivers from the following site for Mac. Since I recently purchased a MacBook, I was in luck. Check out the following site if you are running into this same problem:

http://h10025.www1.hp.com/ewfrf/wc/softwareDownloadIndex?softwareitem=oj-55819-2&lc=ja&cc=jp&os=219&product=1119598&dlc=ja

Good luck!
-swinful

[Solaris] Jumpstart ERROR: (_app_set_slice) Disk is not selected

Today I attempted to leave work early and kicking off my jumpstart installation just before leaving. Waiting to see if the drives would configure before leaving I noticed a failure shortly thereafter:

…edited…
Executing SolStart preinstall phase…
Executing begin script “install_begin”…
Begin script install_begin execution completed.

Processing profile
- Opening Flash archive
- Validating Flash archive
- Selecting all disks
- Configuring boot device
- Configuring SVM State Database Replica on  (c2t0d0s7)
- Configuring SVM State Database Replica on  (c2t1d0s7)
- Configuring SVM Mirror Volume d0 on / (c2t0d0s0)
- Configuring SVM Mirror Volume d0 on  (c2t1d0s0)
- Configuring SVM Mirror Volume d1 on swap (c2t2d0s1)

ERROR: (_app_set_slice) Disk is not selected

Disk layout for selected disks

Disk c2t0d0
Solaris Slice Table
Slice     Start  Cylinder         MB  Preserve  Directory

——-  ——–  ——–   ——–  ——–  ———-
0         0      2061      10241        no  /

2         0     14087      69995        no  overlap

7      2061         2         10        no

Usable space: 69995 MB (14087 cylinders)   Free space: 59745 MB

Disk c2t1d0
Solaris Slice Table
Slice     Start  Cylinder         MB  Preserve  Directory

——-  ——–  ——–   ——–  ——–  ———-
0         0      2061      10241        no

2         0     14087      69995        no  overlap

7      2061         2         10        no

Usable space: 69995 MB (14087 cylinders)   Free space: 59745 MB

ERROR: Flash installation failed

The profiel was correct and looked as follows:

filesys mirror:d0 c2t0d0s0 c2t1d0s0 10240 /
filesys mirror:d1 c2t2d0s1 c2t3d0s1 65536 swap
filesys mirror:d3 c2t0d0s3 c2t1d0s3 8192  /var
filesys mirror:d6 c2t0d0s6 c2t1d0s6 free  /data

metadb c2t0d0s7 size 8192 count 2
metadb c2t1d0s7 size 8192 count 2

After execuring format to make sure the layout was there it dawned on my that at least one of the four drives was previously used for ZFS.  Using ZFS on a whole disk will change the default label on the disk from SMI (UFS) to EFI. Because of this slice-7 could not be seen when performing an install that will use a UFS  layout.

To switch back to SMI I needed to issue “format -e”, then select the disk and “re-label”, choosing [0] for SMI label:

format> label
[0] SMI Label
[1] EFI Label
Specify Label type[1]: 0
Warning: This disk has an EFI label. Changing to SMI label will erase all
current partitions.
Continue? y
Auto configuration via format.dat[no]?
Auto configuration via generic SCSI-2[no]?

Once all the disks were using the EFI label my jumpstart installation gracefully completed.

[Unix] bash$ ssh xmalloc: zero size

On a Solaris 9 system, ran into the following problem when trying to ssh to another host (Solaris 10):

bash$ ssh <host> xmalloc: zero size

Aparently this is a know bug on solaris 9. The quick fix was to: edit /etc/ssh/ssh_config:

and uncomment:

StrictHostKeyChecking no

Thanks to: http://unixtidbits.com/forum/showthread.php?t=1423

[Unix] Solaris: metadb failed to enable/disable SVM service

Noticed the following error while performing a jumpstart install to Solaris 10 (1008-u6).

        - Creating  (c2t1d0s6)

Creating SVM Meta Devices. Please wait …

        - Creating SVM State Replica on disk c2t0d0s7

        - metadb: jade: network/rpc/meta:default: failed to enable/disable SVM s

        - Creating  (c2t1d0s6)

 

Creating SVM Meta Devices. Please wait …

        - Creating SVM State Replica on disk c2t0d0s7

        - metadb: <host>: network/rpc/meta:default: failed to enable/disable SVM…

The installation completed sucessfully, but the error can be somewhat alarming. It is no need for alarm and all it is saying is that the svc service could not be enabled. This is expected as during installation there is no need yet, to enable or disable the svc service for “metadb”.

After installation everything is ok:

bash-3.00# svcs -a | grep meta:default

online         20:24:56 svc:/network/rpc/meta:default

Furthermore:

bash-3.00# metastat | grep State: | grep -v Okay

bash-3.00#

If you are curious my jumpstart profile looks as follows:

filesys mirror:d0 c2t0d0s0 c2t1d0s0 10240 /

filesys mirror:d1 c2t0d0s1 c2t1d0s1 65536 swap

filesys mirror:d3 c2t0d0s3 c2t1d0s3 8192  /var

filesys mirror:d6 c2t0d0s6 c2t1d0s6 free  /data

metadb c2t0d0s7 size 8192 count 2

metadb c2t1d0s7 size 8192 count 2

Oratcl Libs: wrong ELF class: ELFCLASS64

Could not complie Oractl today for the life of me, until a buddy at work pointed out I need to manually edit: generic/oratcl.c to contain:

pt2_obj = Tcl_NewStringObj(“lib32“, -1);
pt3_obj = Tcl_NewStringObj(“libclntsh”SHLIB_SUFFIX, -1);

instead of:

pt2_obj = Tcl_NewStringObj(“lib“, -1);
pt3_obj = Tcl_NewStringObj(“libclntsh”SHLIB_SUFFIX, -1);

The problem was, no mater how I set my environment variable:

setenv LD_LIBRARY_PATH /oracle/10203/lib32:$LD_LIBRARY_PATH
setenv LD_LIBRARY_PATH_64 /oracle/10203/lib:$LD_LIBRARY_PATH_64

running:

tclsh% package require Oratcl

would always produce the following error:

Oratcl_Init(): Failed to load /oracle/10203/lib/libclntsh.so with error ld.so.1: tclsh8.5: fatal: /oracle/10203/lib/libclntsh.so: wrong ELF class: ELFCLASS64

Apparently this has been addressed by a developer and the work around was to (setenv ORACLE_LIBRARY $ORACLE_HOME/lib32/libclntsh.so), http://sourceforge.net/tracker/index.php?func=detail&aid=1525418&group_id=12880&atid=112880.

Oracle made their own ORACLE_LIBRARY environment variable:) What they should have done instead of doing something like:

dlopen (‘/oracle/10203/lib/libclntsh.so’, RTLD_NOW | RTLD_GLOBAL);

done:

dlopen (‘libclntsh.so’, RTLD_NOW | RTLD_GLOBAL);

ThinkPad T400 w/ Ubuntu 9.04-beta

Just installed my ThinkPad T400 using discrete graphics card with Ubuntu 9.04-beta. So far soo good! It just works, fabolous!

I was thinking of returning it for some personal reasons compared to my T43P, but I just might keep it!

Frame Rate w/ Discrete Graphics:

5458 frames in 5.0 seconds = 1090.804 FPS
5841 frames in 5.0 seconds = 1168.091 FPS
5834 frames in 5.0 seconds = 1166.787 FPS
5667 frames in 5.0 seconds = 1133.265 FPS

I am installing WindowsXP to be use side by side. I considerd dual boot but with 4GB of RAM, might as well just keep time side by side and add another 4GB when needed.

So far I am in love with it.

Solaris: Release 10/08 Now Available

Just received an email this morning from Sun that Solaris 10 (10/08) is now available for download. This release includes the following features:

•	ZFS boot on x86 and SPARC systems
•	Support for new Intel Xeon, AMD Opteron, UltraSPARC and SPARC64 processor based systems
•	Visualization enhancements such as Solaris Containers upgrade on attach
•	Networking and security enhancements
•	Device drivers for the latest network cards and other peripherals

I will download this release for a new M4000 I will receiving next month, but will probably give it a run on a desktop nearby to try booting from ZFS!!

AIX: Disk Mirroring w/ mirrorvg

Just last week I received a new IBM pSeries 550! I admit, my first time setting this thing up was a bit frustrating. We did not have any Hardware Management Console (HCM), but through trial and error I was able to get into the console by using a null modem cable and HyperTerminal on a Windows laptop. There are some peculair details regarding the initial setup of the machine, but focusing on disk mirroring I will leave those out for now.

Once the machine was configured and brought on the network. The first thing to do was configure disk mirroring. This is a policy we have in place for all production machines. Drives do fail!! Disk mirroring on AIX, atleast with version 6.1, is pretty easy and straight forward. A lot less commands to type than when disk mirroring with Solaris default software!!

Display volume group information

Since we already have a, Root Volume Group “rootvg”

# lsvg -p rootvg
rootvg:
PV_NAME           PV STATE          TOTAL PPs   FREE PPs    FREE DISTRIBUTION
hdisk0            active            546         488         109..74..87..109..109

for our mirror, all we need to do is extend it onto the second disk “hdisk1″. The first disk is “hdisk0″.

# extendvg rootvg hdisk1
0516-1254 extendvg: Changing the PVID in the ODM.

Checking the rootvg again, we see now that there are two disks on this volume group:

# lsvg -p rootvg
rootvg:
PV_NAME           PV STATE          TOTAL PPs   FREE PPs    FREE DISTRIBUTION
hdisk0            active            546         488         109..74..87..109..109
hdisk1            active            546         546         110..109..109..109..109

Start Mirroring Process

Now that the rootvg is written to the second disk, we can go ahead and start the mirroring process on the second disk, “hdisk1″.

# mirrorvg rootvg hdisk1
0516-1804 chvg: The quorum change takes effect immediately.
0516-1126 mirrorvg: rootvg successfully mirrored, user should perform
bosboot of system to initialize boot records.  Then, user must modify
bootlist to include:  hdisk0 hdisk1.

Taking heed to the message displayed from the completion of “mirrorvg”:

# bosboot -ad /dev/hdisk1
bosboot: Boot image is 36835 512 byte blocks.

With “bosboot” we are saying copy the byte blocks from the first disk “hdisk0″ necessary to boot from the second disk “hdisk1″ should the first drive fail. Looking at the current boot list we just see the first drive:

# bootlist -m normal -o
hdisk0 blv=hd5

The following command configures our boot order to include the second drive after the first:

# bootlist -m normal hdisk0 hdisk1

Verifying we see:

# bootlist -m normal -o
hdisk0 blv=hd5
hdisk1 blv=hd5

FreeBSD: Allow/Configure outgoing mail with Sendmail

If this is the initial configuration of sendmail, then edit:

Edit: /etc/mail/freebsd.submit.mc

repleace, FEATURE(`msp’, `[127.0.0.1]‘)dnl
with, FEATURE(`msp’, `[<ip-of-mail-host>]‘)dnl

or with,  FEATURE(`msp’, `[<domain-name-of-mail-host>]‘)dnl

Then, in /etc/mail, run:

# make all install

Otherwise, edit /etc/mail/submit.cf (about line: 114) to have:

D{MTAHost}[<ip-of-mail-host>]

or

D{MTAHost}<domain-name-of-mail-host>

Neither of these methods should require a restart of sendmail. At least this was the case on my machine.