Monday 4 March 2013

We don't like physically moving servers. Period.

Over the years we have learned may lessons. One of the particularly valuable ones is that we prefer to not move a server once it has been in production.


This particular server is an Intel Server System R2208GZ4GC 2U that has been online for about 90 days.

The server was moved just recently to their new location.

Things went well after firing the server up and bringing services online but today, after being online this last weekend, we came in to some errors.

The status indicator was blinking red. In the Intel Active System Console indicated that we had two error conditions in place:

  • Warning: Memory Channel G1 disabled due to ECC fail.
  • Critical: Blower 2 not present.

After opening the top we could see the above. We re-seated the fan with nothing happening.

So, since we were going to be removing the air duct to get at the memory we went on to check the fan power connector and sure enough it was loose.

We moved the memory stick into another channel to see if we were having a board issue or the memory was faulty.

When we booted the server up we received the three BIOS POST beeps that indicated faulty memory. Sure enough, the little red light followed the problematic stick. But this time we had two fault conditions.

After a few memory channel swaps to figure things out we ended up with the following:


A call into Kingston support and we had an RMA and an advanced swap set up for the problematic sticks. Their support and RMA process has always been consistently good.

In the end it is our preference that once a server has entered service it never gets touched until it is heading for the recycling bin.

Philip Elder
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

Chef de partie in the SMBKitchen
Find out more at

Windows Live Writer

1 comment:

Anonymous said...

Moving servers is never a good idea. The issue worsens in the SMB arena when there may not be a dedicated area / cabinet for the centre of the IT universe i.e. the server(s), switches, routers etc.

We had an instance just 6 months ago where a SMB business owner re-arranged the office and we had to move their SBS server. It is a Friday afternoon of course and we move the server and all seems well until we attempt to boot the server backup, no successful POST.

We had ShadowProtect in place for this client and were able to load up their environment on a workshop loan server whilst we got the mainboard replaced under a NBD vendor warranty. It didn’t cover them for weekends.

Their effective down time was about 2 hours all up. They were very pleased that a) We had implemented a solution which shined in the time of need and b) Their equipment had a Next Business Day response to things were taken care of in a timely manner hardware wise.

Since then they have upgraded to a 4 hour SLA for their server vendor. Overall the server move was a bad idea but it brought positive outcomes for both the business and ours.