WebSphere – The backupConfig script your friend


This is just a short post – But the built-in utility you get with the backupConfig script is worth looking into for everybody!

if you have worked with WebSphere for any significant time you have come across the built-in backup and restore utility that each WebSphere server has by default: [backupconfig.bat] or on Unix/Linux [backupconfig.sh] and the corresponding restore scripts [restoreConfig.bat] and [restoreConfig.sh]. 

At my current client we are working on application customizations and testing them on new servers. This is where the backupConfig comes in handy as it does not just back-up your application server(s) but the deployment manager and all the node(s) configuration as well – so you can replay a whole server configuration along with the installed applications and any application specific configuration. backupConfig can also be used to migrate servers from one piece of hardware to another (or VM server or, …. any combination is possible).

The process is simple: find the scripts in the [/bin] folder of either the Dmgr or the node,  execute the script and it will check your servers configuration, stop all server instances including nodeagents and then create a zip file of ALL files necessary for the back-up – and all of this wonderfulness is  unencumbered by the human thought process … :)

Whenever I am about to install a new application, install any fixes or make configuration changes to a WebSphere  server I run the backupConfig script once first and keep a copy of the zip file it created on a local machine – just to be safe.

Where to run:

Depending on your architecture you can run it on the Deployment Manager and/or all nodes. When you run it on the Deployment manager it will grab all the configuration for the Dmgr, nodes and application servers in one go. This is essential if you have to restore an environment. On a managed node (separate HW) it will capture the configuration and applications installed on that physical node – so you might need to run it on each physical WebSphere server in your environment once to get a total base back-up. Once I have that I usually just run the scripts on the deployment manager as most of the work happens there anyway and all changes are synchronized out to the nodes.

Restores:

On windows it is simple – just run the restoreConfig script and tell it which zip file to use … and presto. On Unix/Linux you have to think a bit more. The backupConfig script does not keep any file rights or ownership information, when restoring it basically sets the file ownership to the account being used to run the script – so make sure you are using the same account and have read/write rights to the folders involved. 

Here is the link to the documentation in the WebSphere Infocenter - I hope you find it useful and make a back-up of your servers soon!

Interesting Tech Note for Linux – Testing the read speed of a hard disk


I look at IBM’s support feeds every day to see what interesting new info comes down the fire hose … this was a first one: Linux drive performance testing! I work with allot of Linux and AIX so this is interesting stuff, I just did not think I would see it in an IBM support feed.

****************************************************

Problem
How can I test to see if all drives are reading at the same speed?
Cause
One reason that hard disks fall out of RAID arrays is a difference in read/write speeds between the drives in the RAID array. You can use the hdparm utility to test each disk to determine whether all disks are reading at the same speed.
Resolving the problem
To run the tests follow the steps below:
Telnet into the server and log in as an administrative user.
Get a list of all hard disks on the system:

wd disk-info

Run the following command on each hard disk on the system:

hdparm -Tt /dev/Disk

Example: hdparm -Tt /dev/sda

The output will look similar to the following:

/dev/sda:
Timing cached reads: 14036 MB in 2.00 seconds = 7027.55 MB/sec
Timing buffered disk reads: 226 MB in 3.02 seconds = 74.79 MB/sec

For hard disks to work together in a RAID, the read speeds should be similar to each other. For best results you should run this test multiple times on each disk and average the results.

Problem
How can I test to see if all drives are reading at the same speed?
Cause
One reason that hard disks fall out of RAID arrays is a difference in read/write speeds between the drives in the RAID array. You can use the hdparm utility to test each disk to determine whether all disks are reading at the same speed.
Resolving the problem
To run the tests follow the steps below:
Telnet into the server and log in as an administrative user.
Get a list of all hard disks on the system:

wd disk-info

Run the following command on each hard disk on the system:

hdparm -Tt /dev/Disk

Example: hdparm -Tt /dev/sda

The output will look similar to the following:

/dev/sda:
Timing cached reads: 14036 MB in 2.00 seconds = 7027.55 MB/sec
Timing buffered disk reads: 226 MB in 3.02 seconds = 74.79 MB/sec

For hard disks to work together in a RAID, the read speeds should be similar to each other. For best results you should run this test multiple times on each disk and average the results.

 

Debmirror – Be independent on Linux


ToalSystems has finally decided that we have enough Linux servers to create out own mirror site – for multiple Linux releases. Over the last week we created a new Ubuntu mirror site on our internal network and will be adding a new CentOS mirror site as well. Currently we are still debating whether we want to register either or both sites for public use as the discussion is still going on whether we want to “adjust/change” the mirror site and add things of our own into it.  If we really decide to make the mirror site public I will blog it on this site.

 

Here is a short primer on how to set up a Ubuntu mirror site using DEBMIRROR – as opposed to RSYNCMIRROR which is a total copy/replica of the site you decide to pull from (including any back-version they have).

It is worth i if you want to have more control over what you offer your Linux  servers/workstations for updates and what versions you support. I also believe in having control where your machines go for an update … I dislike out internal machines going to outside sites to get updates if I can keep them inside. Also, having been burnt by the two big storms we had in the North East (Irene and Alfred) and which severely impacted our data center and the network connectivity, I like not having to be relicant on critical updated having to access the Internet but rather being able to pull all data down from an internal site.

 

 

I am a customer – Connectria


Just a short post for some news. I live in the North East and the winter storm hit us in Connecticut full force and I was out of power for 7 days. along with the power being out for private citizens it also affected most businesses in the state and I lost my mail server for a while.

 

So, I decided to finally create a back-up environment on a waaaaaaaay different part of the grid. So, looking at my options I compared all major ISPs who offer Domino server hosting and after comparing prices and service I decided on Connectria. My old buddy of Idonotes fame  Chris Miller (old as in “I have known him for a while” not as in “he is an old dude”) works there as well so it can’t be a terrible outfit.

 

Turns out their managed hosting solutions were just what I needed and it was rather quick. New CentOS server up, Domino installed and I get root access. Sweet!

 

If you looking to get hosted servers I definitely would suggest you go and check them out, they have a great spread of options to choose from.

Notes on Linux: Run a compact from the OS


I have been pretty busy as of late, hence my reduced blogging – but this is one I come across as a question quite frequently from people who are new to Notes/Domino on Linux.

On the Windows OS it is quite simple to run a compactor task from a command prompt or a batch file. However, when on Linux you will often be met by the following error message:

error while loading shared libraries: libnotes.so: cannot open shared object file: No such file or directory

This error message can appear on either a Linux desktop that has the Notes client installed or a server that has Domino installed. Basically the OS is complaining that it can’t find libnotes.so which usually resides in the /usr/lib folder on your machine. All you have to do is to create some links so that the Notes/Domino code can find this file. And, you will need to do the same for two other files as t the same time.

You will have to be root or have sudo rights to execute this:

sudo ln -s /opt/ibm/lotus/notes/libnotes.so /usr/lib/libnotes.so
sudo ln -s /opt/ibm/lotus/notes/libndgts.so /usr/lib/libndgts.so
sudo ln -s /opt/ibm/lotus/notes/libxmlproc.so /usr/lib/libxmlproc.so

You can see, I have my Notes client (Ubuntu 10.04) installed in [/opt/ibm/lotus/notes] which is the default location. If you have Notes installed someplace else, you will need to adjust the commands above accordingly.

the same commands will solve this issue for Domino server installations as well, though you will likely have Domino installed in [/opt/ibm/lotus/domino]  so just adjust the commands and off you go!

Accountment: Ubuntu 10.04 LTS is now an officially supported platform for Notes 8.5.2


I did just recently rant about IBM being behind on what versions of Linux they support for Notes Domino so this tech note comes timely. I would love to take credit for “pushing IBM along” but  – no matter what truths I twist, not even I can make that case … ;)

The great news is that all us Ubuntu lovers can install Notes on version 10.04 and receive support. Also supported are Red Hat 5.4 (or newer), SUSE Enterprise Desktop 10.0 SP3 or Version 11.0,  and Ubuntu 8.04.

 

IBM Lotus Notes 8.5.2 System Requirements – Linux

 

Domino on Linux: Upgrading to Domino 8.5.2 trouble -> FP are the cluprits


I had been putzing around with my golf swing linux based Domino servers over the weekend. I did run into some issue with upgrading servers but because working my 3/4 pitch swing to the green looked so much more important to me (the weather was GORGEOUS!!) I put it off until later.

Fast forward to today and this is a new technote I just found in the IBM support stream:

IBM – Upgrading from a Notes 8.5.1 fixpack to Notes 8.5.2 for Linux rpm

To summarize quickly: If you have an existing 8.5.1 Domino server on Linux that has ANY FP installed – you will not be able to upgrade to 8.5.2 right away but will have to uninstall the FP to get back to a vintage 8.5.1 install. Once you are back to square eleven you will be able to run your upgrade just as you are used to it. Mind you, you DO NOT have to uninstall 8.5.1, you just have to get rid of your fixpack.

Once the weather goes bad and I don’t feel like working on that golf swing (or my scores on the course nosedive) I will go through the log files on my server I was working on and see if the error messages outlined in this technote match. I want to make sure it was not admin induced error or some other, unrelated issue.

Funny thing is: I have not heard from anybody in my network about this issue yet – has anybody run into it yet?Technorati Tags: , , ,