Andreas' Adventures

Haven't blogged this before. A while ago I posted that my Oracle VM was not working and that I shut it down. Shortly after this I found that the reason was the faulty disk I was using. Rebuild Oracle VM and still happy about it.

When patching the Oracle OIM 10.1.4 to OIM 10.1.4.3 the first configuration assistant might fail immediately. Try starting the listener and issue a retry. This helped in my case.

Very often when you are busy with VMWare, you need an extra disk. Now creating the disk can happen while the VM Linux guest is active. However you don't see the disk immediately. A reboot will help obviously, but I found a nice script on http://www.garloff.de/kurt/linux/ that will show the new SCSI device, so that you can continue and add the disk without a reboot. Just search for the rescan-scsi-bus.sh script. It is on my VMWare Server now and will go directly to each new Linux guest from now on. Any possibility to prevent a reboot is welcome ;-)

Hmm, sometimes life is funny. If I say to you that I have a generic patch, you would expect one patch that can be applied to all versions. Not in this case. I need to apply a patch for the Oracle IDM Suite (10.1.4). Finding the patch is already difficult - and then there are four versions of the same patch - apparently based on different versions of the underlying J2EE. In my opinion this is not generic. Trial and error led me to one of the four versions - which then came back with: OPatch detects your platform as 46 while this patch xxxxxxx supports platforms: 0 (Generic Platform) So again - it seems that my definition of generic differs from Oracle's view. I would understand an error, if the patch was for platform 46 or 18 or whatever. But here a generic patch is apparently not applicable for my platform. Luckily Metalink Note 404473.1 offers a solution. So setting the following: export OPATCH_PLATFORM_ID=0 does trick the opatch into believing that it is no Linux machine any l...

Today I decided to abandon Oracle VM and move back to VMWare. A while ago I bought my old 4GB laptop from my company, when I was given a new one. I added a 1TB disk and installed Oracle VM 2.2 on it. There are some quirks with Oracle VM when it comes to support low end systems. First of all the system is memory bound quite fast. See an earlier post of me (http://achatzia.blogspot.com/2010/05/oracle-vm-memory-issue.html) to see how Oracle VM treats it memory. The second main issue is the problem with external USB2.0 devices. I ended up rebooting an image quite often as the SCSI drive was declared dead. I worked with VMWare since about 6 years, and only had lost one image - a colleague disconnected my external drive which resulted in a corrupt Oracle database. With Oracle VM I never had that stability. So for me it is good-bye OVM, and welcome back VMWare Server. Obviously the VMWare Server will run on a Linux host with Oracle Enterprise Linux 5.5 .

I was reinstalling my Oracle VM (again). During the configuration of the Oracle VM Manager template I ran into an error: OVM-2007. The template (new OVM image) could not find the host it was running on. Quick check in the /etc/hosts file of the OVM Server showed: 127.0.0.1 ovm.mydomain.local ovm localhost.localdomain localhost So I changed it to read: 127.0.0.1 localhost.localdomain localhost 10.1.1.100 ovm.mydomain.local ovm localhost.localdomain localhost To me as a Unix veteran it is still a mistery why somebody came up with the idea to put anything besides 127.0.0.1 localhost into the first line.

Recently I had a lot of disk errors on my Oracle VM server. The errors indicated that some scsi problems were manifesting itself with: unable to enumerate USB device on port 2 After unloading and reloading the ehci_hcd it seems to be gone.

Today I tried to cut some corners and use a silent.xml file for WLS 10.3.3 for a silent installation of 10.3.2. To my amazement on a clean machine I received the following error: oracle@xxx001:/opt/oracle$ java -Djava.io.tmpdir=/opt/oracle/tmp -jar /nfsstage/wls1032_generic.jar -mode=silent -silent_xml=./silent/wls_silent.xml -log=/tmp/wls.log Extracting 0%....................................................................................................100% The local BEA product registry is corrupted. Please select another Middleware Home or contact Oracle Support I checked the wls_silent.xml file and found that it includes Coherence, which was not bundled with the 10.3.2 version. So I modified the 10.3.3 silent_xml file from value="WebLogic Server/Core Application Server|WebLogic Server/Administration Console|WebLogic Server/Configuration Wizard and Upgrade Framework|WebLogic Server/Web 2.0 HTTP Pub-Sub Server|WebLogic Server/WebLogic JDBC Drivers|WebLogic Server/Third ...

Nobody likes to write documentation, and I'm not that different. But today I was very happy and said "Thank You" to myself. A while ago I needed to install some software for a project. Now after my vacation, which apparently wiped a lot of my mind's hard disk clean, I needed to repeat the installation for a new environment. Reverting back to my own documentation saved me a lot of time. Maybe I should create a video on YouTube with my impersonation of Steven Balmer's monkey dance and substitute the words "Developers, developers..." with "Document! Document! Document!". So take it from me: writing documentation before your vacation keeps you longer in your post-vacation mode ;-)

A few posts ago I mentioned that the silent install was very slow. Still had this on my to-blog list, so here is the clarification. It turned out that somebody had the idea that we needed some kind of virus protection on the Linux boxes. Each time something was written to the disk the virus protection kicked in. Imagine what this is doing to your installation performance :-(

When using the Oracle Database for the storage of WebLogic Portal settings you need to modify the database.properties and run the create_db.sh script. In the database.properties the example uses the Pointbase, so the values in the Oacle section are filled with placeholders. When replacing the jdbc string be carefull not to remove all @ signs. The original value is: oracle.url=jdbc:oracle:thin:@@DB_HOST@:@DB_PORT@:@DB_NAME@ The fisrt @ after the 'thin:@' needs to remain there. Otherwise you will get a sql exception when you run the create_db.sh script.

It seems that I'm all into silent installs these days. Trying to do a WebLogic installation in silent mode. Went to the documentation, as a few things have changed since the early BEA days. So I found the sample file for the silent.xml . Without big thinking I copied the sample like this: &ltbea-installer&gt &ltinput-fields&gt &ltdata-value value="D:\Oracle\Middleware_Home" name="BEAHOME"&gt &ltdata-value value="D:\Oracle\Middleware_Home\wlserver_10.3" name="WLS_INSTALL_DIR"&gt &ltdata-value value="WebLogic Server/Core Application ServerWebLogic Server /Administration ConsoleWebLogic Server/Configuration Wizard and Upgrade FrameworkWebLogic Server/Web 2.0 HTTP Pub-Sub ServerWebLogic Server/WebLogic JDBC DriversWebLogic Server/Third Party JDBC DriversWebLogic Server /WebLogic Server ClientsWebLogic Server/WebLogic Web Server Plugins WebLogic Server/UDDI and Xquery SupportWebLogic Server/Server Example...

Yesterday I did my first ever silent install. Nothing fancy yet. There are two possibilities: a) Silent install is dead slow or b) There is something wrong with the machine or the storage Bottom line is, it took me 1.5 hours to just install the database software. I need to investigate this, as this is not what I had in mind. Continuing now with the listener and the dbca in a silent manner.

... or Google Chrome has a problem with counting ;-)

I finally have made the jump over to Australia. After a somewhat bumpy first week (oh man - what a jet-lag) I started to settle down. On my project I will do a lot of stuff with Coherence, WebLogic Portal, the OSB, Grid Control and a RAC database. I'm looking forward to do this. This is part of re-inventing myself, while being in a nice place. I will also try to blog some of my experiences in http://yetanotheraustraliablog.blogspot.com/

After some first baby steps with Oracle VM I wanted to push this thing a little bit further. So my idea was to have more than a couple of VM's running simultaneously. The Oracle VM told me however that I had not enough memory to do this. So I was sitting in front of my dedicated laptop and told the screen that there should be 4 GB inside. The laptop however had one of his funny days and told me (using free and top) that really nothing more than 572 MB would be inside the machine. Hmm - when a machine tells me something I am the first to listen. So I shutdown everything, got my screwdriver, opened the memory compartment and had a look. Indeed 1 bank of 2GB. I didn't check the second bank as this is hidden underneath the keyboard. Closing the machine and powering it up, I prepared myself for another round of telling it that it should recognize a little bit more than 572 MB. I examined the dmesg and still it told me that there was only 572 MB inside. Now imagine, you have just see...

Another thing I came across when dealing with the OSB was that the Workshop throw an error and didn't start at all. When you end up with a large dialog box showing a list of startup parameters (which all originate from the workshop.ini file) have a look at this site: http://wiki.oracle.com/page/How+to+get+Oracle+Workshop+for+Weblogic+10GR3+working+in+Linux?t=anon

During the startup of the Eclipse Workshop in my OSB environment I encountered the following error: “requested array is larger than heap” The solution (from the OTN forums) is to delete the publish*.dat from the following directory: FMW_HOME/user_projects/workspaces/YOUR_WORKSPACE/.metadata/.plugins/org.eclipse.wst.server.core/publish/

I am busy installing the latest version of the Oracle Enterprise Manager. Obviously I need a database for this. In order to protect my resources in the Oracle VM image that use I switched off all extra Database options. Among others the Partitioning option is offered. Well - during the installation of the OEM I found out the hard way that the Partitioning option is required for the database. I know that the OEM creates a lot of partitions - which will then get emptied during the data consolidation process. I happened to forget this. The funny thing is that the OEM installer checks this and comes back with a warning box, stating: Using the Partitoning option is strongly recommended. Being a wise guy I tried to ignore this. But when Oracle says that they STRONGLY RECOMMEND something they mean it. So I am deinstalling and reinstalling the database software as I write this ;-)

Just for your convenience: When you start an Oracle VM image from the commandline with xm create -c ./vm.cfg you will see that the boot process continues until the login is displayed: EXT3-fs: mounted filesystem with ordered data mode. type=1403 audit(1272761495.554:2): policy loaded auid=4294967295 ses=4294967295 Enterprise Linux Enterprise Linux Server release 5.4 (Carthage) Kernel 2.6.18-164.0.0.0.1.el5xen on an i686 soa.mydomain.local login: In order to get back to your commandline type CTRL-] and you will get back the control.