OBIEE 11.1.1.6.2 BP1 SampleApp v207 on Virtual Box 4.2.12


You might have read of how long and complex it is to install OBIEE directly onto Redhat.

OBIEE SampleApp v207 Desktop

OBIEE SampleApp v207 Desktop

This year I heard that there was a new Virtual Box image of OBIEE. The idea being: you install and start Virtual Box, then use the OBIEE virtual machine. Great idea. You should be able to get on with actually using OBIEE, not struggling with its installation, startup, and administration.

However, that would assume that all the components work. Like any technology stack.

Unfortunately for me, they did not.

Installing Virtual Box 4.2.12 r84980:   

My previous experiences with Virtual Box 4.1-4 were not good. First, installing it on Redhat 5.5 was a project. Then, it completely hung my Redhat server!

As you can imagine, after that experience, I am rather leary of Virtual Box. But hey, maybe it’s improved. And, on Redhat 5.5, I’ve got Oracle Applications R12 and OBIEE. Why not just add another hard disk to my server, and dedicate it to OBIEE? If it screws up, I’m not affecting the other apps, and I can just overwrite the disk, right.

In the end, I test drove the Virtual Box and the OBIEE VM on both Windows 7, and Redhat 6.4 64 bit host machines.

On Windows 7 64 bit, I downloaded the (32-bit/64-bit)  VirtualBox-4.2.10-84105-Win.exe file from:
http://www.oracle.com/technetwork/server-storage/virtualbox/downloads/index.html

Last time on Redhat, I installed Virtual Box by downloading and installing an RPM file. This time, I used yum to download and install it. I used these instructions:
http://www.if-not-true-then-false.com/2010/install-virtualbox-with-yum-on-fedora-centos-red-hat-rhel/

Virtual Box 4.2.12 r84980 actually installed cleanly enough on both operating systems. Hurray!!!!  This is a vast improvement over my previous experience. Did the engineers read my previous posts?

However, the OBIEE VM was another story.

Unzipping The OBIEE Image:

I downloaded the OBIEE image from:
http://www.oracle.com/technetwork/middleware/bi-foundation/obiee-samples-167534.html

It turns out that the OBIEE SampleAppv207_OBI_BP1 is huge. About 22 gigs. It contains the Redhat operating system, the Oracle database, and all the OBIEE components to go on top of it.

It’s supposed to be a great big single VM image. To download, it’s split up into eleven 2 gig files.

SampleAppv207GA_OBI_BP1.zip.001

SampleAppv207GA_OBI_BP1.zip.010
SampleAppv207GA_OBI_BP1.zip.011

There is also the VB Image Key File: SampleAppv207_OBI_BP1.ovf.

You are then supposed to then zip them back up into the single 22 gig image. But there are no instructions on how to do this. Not even in their manual.

I put out some questions to newsgroups, including Oracle, but no one knew. Not even the guy who wrote the 7-zip software had any idea how to do it on the command line!

So, here is the secret. After you have downloaded the files, you don’t use the command line to unzip them as you would expect. You use the GUIs: 7-ZIP or the Redhat file manager to do it.

In Redhat, from my notes: with file manager, right click the file SampleAppv207GA_OBI_BP1.zip.001. It shows the two big vmdk files. Extract to: /obiee_vm
The Redhat (un)zipping software was File Roller 2.28.2.

In Windows, right click the file: SampleAppv207GA_OBI_BP1.zip.001. Use 7-zip to extract the archive.  So, on Windows, you have another software as a prerequisite: 7-zip. The unzip process will create two files:
SampleAppv207_OBI_BP1-disk1.vmdk –  22G
SampleAppv207_OBI_BP1-disk2.vmdk – 1.5M

OBIEE 11.1.1.6.2 BP1 SampleApp v207:

Unfortunately, neither the Windows, or the Redhat sample app worked well enough to actually be used.

They were total pigs on the CPUs and glacially slow. When I say slow, I mean it would take over a minute to change focus from one terminal window to another!

After starting just a few OBIEE scripts, I gave up on Redhat. It was obvious that it was not going to work.

On Windows, once, I did manage to start all the scripts. However, it took about 2 hours to actually get all of them started! Then the next time that I tried, after engaging the third startup script, StartWLS,  the weblogic server never did start. But all the CPUs continued to go crazy. Most well over 90%.

OBIEE Virtual Machine Info:

Once you start the OBIEE VM, it boots a Redhat 5.8 virtual machine. Here are some key stats:

cat /etc/redhat-release
Red Hat Enterprise Linux Server release 5.8 (Tikanga)

uname -a
Linux obieesampleapp.us.oracle.com 2.6.18-308.1.1.0.1.el5PAE #1 SMP Wed Mar 7 11:42:30 EST 2012 i686 i686 i386 GNU/Linux

lsb_release -a

LSB Version:    :core-4.0-ia32:core-4.0-noarch:graphics-4.0-ia32:graphics-4.0-noarch:printing-4.0-ia32:printing-4.0-noarch
Distributor ID: EnterpriseEnterpriseServer
Description:    Enterprise Linux Enterprise Linux Server release 5.8 (Carthage)
Release:        5.8
Codename:       Carthage

hostname
obieesampleapp.us.oracle.com

domainname
(none)

getconf -a | grep -i “long_bit”
LONG_BIT                           32

It’s actually a 32 bit virtual machine, on 64 bit host machine.

Makewhatis:

One very strange thing. Once the Redhat virtual machine started, makewhatis engaged and was seen in top. Some investigation revealed that these processes are actually set in cron jobs! And, not one weekly cron job, but another daily cron job too!

/etc/cron.daily/makewhatis.cron
/etc/cron.weekly/makewhatis.cron

This is totally bizarre! Why would you ever put makewhatis into a cron job? I’ve always run it manually. Perhaps because of updates, you might want to run it periodically. But why not just run makewhatis manually, after you manually clicking the button to allow the updates?

Not only that, but whoever wrote the cron jobs, was actually concerned with both jobs running at once! From the file:

cat /etc/cron.weekly/makewhatis.cron
#!/bin/bash

LOCKFILE=/var/lock/makewhatis.lock

# the lockfile is not meant to be perfect, it’s just in case the
# two makewhatis cron scripts get run close to each other to keep
# them from stepping on each other’s toes. The worst that will
# happen is that they will temporarily corrupt the database…
[ -f $LOCKFILE ] && exit 0
trap “{ rm -f $LOCKFILE; exit 255; }” EXIT
touch $LOCKFILE
makewhatis -w
exit 0

And, if you think of it, if you are running the job daily, why would you also run it weekly? Imagine cutting the lawn, would you also have a weekly and daily schedule? At 10am, the daily guy cuts the lawn. At 2 pm, the weekly guy cuts it. Duh!

Yum Updates:

If that is not enough, the yum updates are configured to run once an hour, and multiple configuration files!

cat /etc/yum/yum-updatesd.conf
[main]
# how often to check for new updates (in seconds)
run_interval = 3600
# how often to allow checking on request (in seconds)
updaterefresh = 600

find /etc   -type f | grep -i “update”

/etc/dbus-1/system.d/yum-updatesd.conf
/etc/rc.d/init.d/yum-updatesd
/etc/updatedb.conf
/etc/yum/yum-updatesd.conf

I don’t think that even Windows anti-virus software checks for updates as often!

Windows 7 Host Results:

The one time that I got all the scripts to start. But even so, the end result was not successful; I couldn’t see any results in the firefox browser:  Firefox can’t establish a connection at localhost:7001

9:32 am boot VM

9:35: gets past udev

9:36:  get Redhat login screen, login

9:37 – desktop appears

OBIEE 11.1.1.6.2 BP1 SampleApp v207

I had allotted only 4 CPUs, system manager shows only 4 CPUs running.  In contrast to Linux, which used all CPUs, no matter what the settings were.

yum updates engage.
kill them

9:42 – start database.
It starts right away

9:43:  1 – start OID

CPU goes up.
A few seconds slowness in highlight and right click.

In the VM, the CPUs all go up to over 90%.

ias-component/process-type/process-set:
oid1/oidmon/OID/

Error
–> Process (index=1,uid=968250563,pid=3955)
time out while waiting for a managed process to start
Log:
/home/oracle/oid/bioid2/diagnostics/logs/OID/oid1/console~OID~1.log

9:55: CPUs finally settle down.

9:57: 2 – start WLS
weblogic

Some CPUs go over 90%.

10:07: get to
<BEA-000365> <Server state changed to STARTING>

10:12:  <Server started in RUNNING mode>

15 minutes to start weblogic!

CPU comes down.

10:13: 3: StartB1.sh

10:24:  finally starts

9 minutes to StartB1

—-

10:26: 4 – startEssbase.sh

10:27: seems started, no more terminal window.

10:28  5 – startEssbAPS.sh

10:37  – finishes

9 minutes to start the script: startEssbAPS.sh

10:40: 6-StartEAS.sh

10:57: finishes

17 minutes for StartEAS.sh!

—>   Total time: about 1 1/2 hours to start all the scripts!  <———

OBIEE VM, all CPU Near 100%

OBIEE VM, all CPU Near 100%

Then, all the VM CPUs were very close to 100%!

Bring up Firefox. Click applications, internet, firefox. After about 2 minutes, finally get Firefox.

But: Firefox can’t establish a connection at localhost:7001

netstat -a | grep -i “7001”
– nothing

try in the browser: localhost:7001

Firefox can’t establish a connection at localhost:7001

—-

About 2 hours, just to get started. And it’s still not working!

——–

Go to shut it down.
How long will that take?

11:27
1-StopBI.sh

opmnctl stopall: stopping opmn and all managed processes…
opmnctl status: opmn is not running.
Hit return to continue

11:28
2-StopWLS.sh

– stops in about 1/2 minute.

11:29

3-StopEssbAPS.sh
– stops right away.

11:30

4-StopEAS.sh

This brings up all the CPUs over 90%.

11:31 – script window disappears.
But all CPU are still going close to 100%, then die.

11:32
5-StopEssbase.sh
– stops in about 20 seconds.

11:33
6-StopOID.sh

11:34

7-StopDB.sh
– stops in about 20 seconds

11:35
8-StopTT.sh

– stops in about 10 seconds.
TimesTen Daemon stopped.
Hit return to continue

11:36
– shutdown linux.

About 8 minutes to shutdown the stack.

(after over 1 1/2 hours to start it! )

——

Windows, Round 2:

I increased the number of CPUs to 6 (of 8) and the memory to 11 (of 16) gigs.

However, this time, the OBIEE stack hung after I started the third script: 2-StartWLS.sh.  The whole process hung, and all 6 CPUs were totally engaged.

OBIEE VM, After Starting Weblogic

OBIEE VM, After Starting Weblogic

At this point, it habitually took 35 seconds just to change the focus from a terminal window to System Monitor!  I gave up.

Results On Redhat Host:

The results on the Redhat host were even worse than on Windows.

OBIEE VM on Redhat Host

OBIEE VM on Redhat Host. Massive CPU consumption With No Load

/usr/bin/virtualbox &

9:00 pm, Host is up and running fine.
Start Virtual Box.

In System monitor, Virtual box is taking up about 60% on all 8 CPUs

top: 455%
(8 cpus on the machine.  So, 455/800)

9:03 pm
– still at Starting UDEV

Can move around the Host OS ok.
Take screen shots.

9:06
– still at Starting UDEV

9:07 – begins to move on.

“Setting hostname”
boot proceeds more normally, but slower than usual.

Starting Times Ten Daemon

on host, top indicates:  455% being used by Virtual Box

9:10
Starting Unbreakable Linux Support
Login to Oracle

9:11 black screen

9:13 – get VB guest screen.

It takes 13 minutes just to boot the Redhat VM!

—-

9:16
host:  top:  400%

VM:
Go into system monitor.
Try to change some colors.
A few second delayed reaction to moving the slider.

9:21

Double click the terminal icon.
Takes 33 seconds to bring the terminal up!

Takes 2 seconds to change focus from one window to another.

ps -ef | grep -i “yum”

root      3704     1  0 19:09 ?        00:00:09 /usr/bin/python -tt /usr/sbin/yum-updatesd
root      4084  3704 31 19:12 ?        00:04:41 /usr/bin/python -tt /usr/libexec/yum-updatesd-helper –check –dbus

ksoftirqd on top: running up to 70%

To change focus from one window to another takes 4 seconds!
Another time: 10 seconds!

To maximize System Monitor took about 10 seconds!

Start the database.
Takes about 6 minutes to start!
During that time, changing window focus is very slow.

After that, to change focus from a terminal windows to System Monitor,
so that all the cells in System Monitor fill in with color, takes 55 seconds!

To change focus back to a terminal window running top:  10 seconds

To change focus back to System Monitor, about 1 minute, 15 seconds!

System Monitor normally runs at about 1 second or less update speed.
Now it is taking a good 5 seconds to see the numbers update.

Take a break.

12 midnight
VM top: an Oracle process running at 49 to 67%

At least 9 “oracle” processes running.

top: in the top 10 processes, 6 are “oracle”.

time ps -ef | grep -i “oracle”

real:  16.2 seconds.
user:  .94
sys:   11.3 seconds

However, it takes over a minute in actual time.

Much of the time, the mouse disappears in the VM. I can see it when I’m navigating in the host. But not the VM.

In the VM, in a terminal window, hit return three times. It takes 20 seconds to give the first response from hitting the return key. 30 seconds to leave three more blank lines.

time ps -ef | grep -i “oracle”

Actual time: 2 minutes, 35 seconds for the command to complete.

real:    32.0 seconds
user:    .9
sys:      16.8 seconds

So, where did the other two minutes go?

16 CPU seconds, just to get a process list!

time ps -ef | grep -i “orcl”

1 minute 20 seconds to finish.
22 actual seconds, just to clear the return and get a new line.

real:    21.0 seconds
user:    .9
sys:      11.8 seconds

run the command:
which oracle

it takes 30 seconds to finish.

Highlight, right click to copy.
It takes about 30 seconds to bring up the popup menu to copy.
But I can’t get the copy to work, to bring the data back to the host.

time which oracle

Takes 35 actual seconds to finish

real:    10.8 seconds
user:    .4
sys:      7.9 seconds

1:25 am

Can’t get to “system” to shut down.
Finally:
init 0

Thoughts:

In 2010, I did see the OBIEE VM in use at Oracle Days, and it seemed to work fine. On the attendee’s laptops! So I’m rather perplexed why the latest version is so unusuable.

I can speculate or investigate this or that. Perhaps because I’m using a 64 bit server, and the OBIEE VM is 32 bit. I could try to reinstall. Or debug it.

However, I’ve already spent dozens of hours over many weeks trying to get it to work. How many more hours, over how many days or weeks should I spend struggling with to get the OBIEE VM to work? I’ve already got OBIEE running on my Redhat 5.5 server.

If anyone has done this installation, and it worked normally, please detail the technology stack and what you did. If you had the same runaway CPU symptoms, and figured out how to fix them, please let us know to help any other poor soul.

Without starting the OBIEE appliance, the VM is usuable. Perhaps it could be used for something else. However, why not just install Redhat in a new VM, and do what you need?

Conclusion:

Virtual Box has improved since the last time I worked with it. It actually installs cleanly on both Redhat 6.4 and Windows 7. Great!

However, even on my dual Xeon server with 16 gigs of RAM, the OBIEE 11.1.1.6.2 BP1 SampleApp v207 on Virtual Box was totally unusuable.

Advertisements

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: