After a few hours of diagnostics and upgrades I finally figured out what the issue was, the SuPHP module was attempting to run php5 on the server, unfortunately php5 was a symlink to the php-cli executable and not the php-cgi executable as it was expecting.

The fix was simple in the end, issue the following command.

ln –s /usr/local/php /usr/local/php5

Hope this saves someone else some head scratching!

At the present time I’m using Microsoft One Note 2007 to compile my notes, its working out quite well at the moment.

My aim is to pass this exam and then begin down two routes, the first will be to upgrade to MSCITP: System Administrator and the second is to pass the SharePoint 2007 related exams as we use SharePoint quite heavily within the company I work for.

Now that Windows 7 has reached our MSDN account I decided it was time to test it properly, downloading the 32 bit version of Windows 7 Enterprise Edition I began the installation. Disheartened by two failed installs due to bad media (it appears my laptops DVD-RAM drive burns coasters!) I finally got the OS installed.

My initial impression’s of the install process and the system were good, the install took a grand total of 7 clicks to complete, the only slow point was the expansion of the files on disk and the initial configuration. Despite the slow speed it only took 2 hours to install, the first boot time was impressive. It takes about 30 seconds to boot to a usable desktop which doesn’t hang on me when I click and ask the machine to perform a task.

I did a quick scout around the internet for information on making Windows 7 more SSD friendly, I was pleasently surprised to find that the OS had carried out most of the tweaks suggested automatically as it knew it was installed on an SSD – thats a huge step forward from the RC we tried out in the office.

The only downsides so far are the terrible shutdown speed of the machine and the time taken to install Windows Updates – those though are things I can live with for now.

Following the advice given to us by a 3Com engineer who had been on site a few months earlier we decided to route the new Fibre Connection through our core switches a nice set of 5500G-EI 48 Port PoE Switches. Doing as I’d been taught I created a VLan on the switch as it didn’t have a IP Address it wasn’t routable accross the switch. This VLan was then plugged in to the Fibre Converter and our firewall, allowing us to place devices outside our firewall (Which we had wanted to do previously with a few of our servers which have issues working through our ISA Firewall).

That portion of the work went swimmingly, however the next morning as I sat down to document this change I was told by the boss – “We don’t have the time, there’s no money in the budget to write it up. Anyway it’s not somehting you’ll forget now is it?”

Fast forward 9 months and whilst on annual leave I get a phone call which went somethng like this.

“It’s all dead, we have no connectivity with the outside world.”

“Oh, well have you checked the connections?”

“Well I would, but I don’t know how its all plugged together.”

“Ah” I responded “I know I did the work but I can’t remember either.”

Cue me kicking myself as thoughts of being dragged in to the office emerged in my mind, after an hour of phone diagnostics we finally figured out A) how it all hangs together and B) what exactly the fault was. The very next day I was back in the office and a sitting at my desk, madly typing up a document describing how things are installed and setup as I really don’t want another phone call when I’m on leave.

The reason for typing this out is two fold, one to hopefully educate others on the importance of documenting your work and also to educate myself on the importance of documenting as I do things – even if there isn’t the budget for it. A job that would have taken me 2 hours the day after setting up the network took me a whole day 9 months down the line.

Saying we don’t have the time creates a false enconomy, the question you should be asking is can we afford the time it will cost if we don’t document?

~Dan

http://java.sys-con.com/node/1027624

~Dan

In testing we noticed an issue where by if you tried to deploy the Ghost Client to a PC in a workgroup it would fail with the following error message.

Failed to install:
	\\PC780
	(7): Connection: Logon failure. Possible reasons are:
		1. Incorrectly specified password and/or username.
		2. Simple File Sharing is enabled and the client is running on Windows XP.
		3. Incompatible Windows LAN Manager authentication levels between console & client machine.
		Please refer to the Symantec Ghost Manual to change the above mentioned windows settings.

The fix was to disable Simple File Sharing on the target PC.

This can be accomplished by navigating to the tools menu in Windows Explorer, choosing Options then clicking on the view tab. The last option on that window is to Enable or Disable Simple File Sharing.

Simply untick that box, and click okay.

You will then be able to install the Ghost Client.

What followed was an issue where the software refused to load Steam.dll.

The issue was related to the fact the corporate firewall was blocking steams port, in my case I added the machine to the CMP allow rule which resolved the issue.

In your environment you may need to figure out exactly which ports to open.

~Dan

It’s just a pity that we rolled out the software nearly 6 months ago, and that I finished testing about that long ago. There was simply a scheduled wipe still in the system, which got triggered last night when I tried to sync my handset again for the first time in ages.

All my numbers have gone because I didn’t have a recent backup of that device.

Oh well, its a lesson learned.

When you’re finished testing always check you’ve undone everything you messed around with – better still start again from scratch in case you’ve missed anything.

~Dan

In addition to enabling wol in the bios you also have to set the power save mode to S3 and ensure that low power mode is switched off.

Hope this saves someone considerable time!

~Dan

The company I work for (like many other companies of its ilk) is a little strange, born out of academia it has always done things the SCI way whether that way conforms with industry best practice or not. Often in the past it has done things under the impression it would change best practice by doing the task in a new manner.

One such issue has been the companies IT systems.

The root causes of the issue are

1. We’re an educational establishment, we get cheap software.
2. We’re not for profit, so we tend to spend the excess on shiney new stuff.

What this lead to was a proliferation of servers over the past 8 years, every new project lead to us provisioning a new server.  At the time you never noticed the effect the growth was having on the IT department, for the few few years of their life servers generally don’t cause much fuss – its only when they get old they start missbehaving.

The cheap cost of software and its almost instantanious availability (Microsoft Licensing rocks!) meant that we have always been able to play with toys that would normally be outside our reach, for instance we have gone through 4 versions of exchange in the last 6 years who else would have the money to do this? I know several companies that are still standardised on Exchange 5.5 because it works and does the job well.

Along side core systems like Exchange we run several special services, such as Microsoft Content Management Server and SharePoint. Since I joined the company (3 years ago) we’ve gone through 3 versions of SharePont server alone.

Now, Evan asked how we can possibly have 40 servers. I will now try (as generically as possible) to explain it.

Internal Domain

Firstly we have 3 domain controllers supporting our internal domain, this follows a disasterous couple of weeks this year when we lost both of our previous domain controllers when their virtual hosts suffered issues. This prompted us to move the servers back to hardware, just to be sure that if hardware issues do strike again they don’t take down everything, that outage really highlighted what the loss of our internal Domain does to certain web services we run that authenticate against it.

We also run Exchange 2007, currently on a blade supported by an AX150i SAN. Along with another server which acts as a mail relay and anti spam Edge server. We are also running an Exchange 2003 internally for the next few months as we finalise our migration to 2007.

For our file storage needs we run two identical NAS devices, these are set to replicate data between the master and the slave to ensure if the main server goes down that we can switch to a current backup.

Count so far 8.

We also run an app server, this box runs software such as AutoCAD which need to be installed on a central server and deployed out to clients.

Our business system runs on another server, co-existing Navision with other software is a bad idea, we know this because we tried it.

Ansys Server we maintain one server dedicated to our engineering modling software, that box is bascially there to number crunch for us.

Although its virtualised, our print services count towards the totals.

Count so far 12.

After our foray in to Virtualisation we have 4 servers dedicated to this task, two run Virtuozzo Server, one runs Micrsoft Virtual Server 2005 and the latest run is running Hyper-V.

Count So far 16.

Perimeter Domain

Moving on to our perimeter domain we start again with domain controllers, this time two. One hardware and one virtualised box.

DNS server. We host a fair number of DNS records, after several years of issues running DNS with another service we decided to split it off to its own virtual instance.

Exchange 2003. We maintain a second Exchange server on the perimeter domain, this is for use by our clients and the mailing systems attached to our websites. We have tried other mail servers for this role but only Exchange Server has given us proven reliability.

Database Servers. We are in the process of migrating from SQL Server 2000 to SQL Server 2005 at our DBA’s request we (willingly!) avoided an in place upgrade of our server and provisioned a second box to do the job. Until the migration is complete we’re running two SQL servers.

Count so far 21.

IIS. We run a server dedicated to IIS which at the present time hosts around 40 sites of various ages.

Microsoft Content Management Server. A few years ago we invested time in MCMS we now maintain a server running 5 MCMS sites which we will be looking to maintain for the next 3 to 5 years.

SharePoint 2003. We have clients running extranets on this software, its another box.

SharePoint 2007. Installed last year we are now using SharePoint 2007 for our new sites and extranets.

Issue Tracker. Due to security issues with our issue tracking software it exists in its own virtual instance.

Count so far 26.

Bookshop. We have a very old server running our bookshop, installed in the year 2000 and never touched again.

Ultraseek. Our enterprise search service supports three of our frontline servers.

Other Servers

Firewall. Kind of important, we have a box dedicated to routing / VPN / Firewall duties.

Development Servers. We are running 5 development servers of various ages to support the web developers.

Another App Server. We are running a second app server under virtualisation because it won’t run on our desktops. Another to add to the mix.

The CMP server. My baby, recently retasked from another less glamerous life it now runs the management software I have been clamouring for over the last 3 years. Micrsoft Systems Center Operations Manager 2007, Configuration Manager 2007 R2 and 3Com Network Director.

Final Count 36.

Wait. Our VOIP system that got forgotten.

Final Count 37.

Okay, so 37 is a little lower than I originally stated but that is only because I managed to remove a few virtual servers this week by being ruthless.

Is 37 too many for one person to manage? Probably, as you’ll see above I forgot one again. Time to print this entry and put it on my wall just incase I forget again.

~Dan