Video Screencast Help
Symantec to Separate Into Two Focused, Industry-Leading Technology Companies. Learn more.

Ghost installation brought our whole company down for half a day

Created: 19 Mar 2010 • Updated: 22 Sep 2010 | 1 comment

I downloaded and installed the Ghost trial, as our reseller assured me that I would only need to install the license file into the trial version to make the trial version a live version.  After I received the license file, I could not find a way to install it, and was subsequently advised by you guys that no, I would need to uninstall the trial version and install instead a live version.  I did that, and the live version cleared up the issues that I was experiencing with creating images as well.

Our SQL servers use a login account (user ID) that I will call sqlsrv.  The server that I installed Ghost on used to be a SQL server, so it was logged in under the sqlsrv account.  As directed, I uninstalled the trial and installed the live version, on Wednesday.  On Thursday people that use our ERP system - the whole company - started experiencing various odd errors.  People could not save sales orders, etc.  It got worse and worse, so I finally got everyone in the company out and rebooted the SQL server.  When it came back up, it could not use the sqlsrv account.  It logged in and created a new sqlsrv.xxxx account, where xxxx is our domain name.  After digging for a while, we found that the mailbox for the account sqlsrv had somehow been deleted.  This is the cause of at least the problem with saving sales orders; there are many triggers that fire off automatic e-mails if certain conditions are reached, like if the sales order exceeds the customer's credit limit, an e-mail goes to the credit manager, etc.  Since the mailbox no longer existed, it caused the error.  And there were other errors.  Bottom line, we could no longer use the sqlsrv login, and other things were tied to this login.  Figuring out that  this was the problem and setting up a new login and profile and bring the SQL server up under the new profile took half a day, and the server still isn't running correctly.  We are, to put it very mildly, VERY unhappy that installing Ghost as directed, which meant having to uninstall the trial first, cause the profile to go south and so subsequently caused our main live SQL server to crash. 

Should Ghost have been installed upon a server that had the same login profile that another server also used?  In retrospect, clearly not, but this was nothing that we would have realized before the error occurred.  Nothing in the documentation that I have been able to find that indicated that uninstalling the trial, to then install the live software, would do anything to the servers login profile, delete the mailbox, etc. 

In the trial Ghost, I also logged into the machine that I was trying to create an image of with an internal utility login, itdev, that we use for general login purposes within the IT dept.  After working with Ghost for a while, I decided to instead create a new user ID/profile that would only be used for Ghost, so I rebooted the source computer and logged in under this new login ID and kept testing with it.  We also discovered last night that other computers, which run diagnostic and monitoring tools on the network, have all failed, as the ITDEV login appears to have also been corrupted and no computers can log in under this account now.  My guess is that somehow switching the user ID on the source computer and then subsequently uninstalling the trial version somehow corrupted the itdev user profile, causing additional very substantial problems that we are still trying to recover from.

A virus could not have done more damage to us than Ghost did, and there was nothing anywhere that I could see that hinted at anything like this being a possibility when using Ghost.  What a disaster this has been.  

Comments 1 CommentJump to latest comment

EdT's picture

Never trial software on a production system. You unfortunately found out the hard way that this is never a good idea.
Test it on a suitable test environment that mimics your production environment first, and iron out the bugs in there rather than on live boxes.
Consider yourself lucky that it only took half a day to recover a semblance of normal operation - viruses can and have caused much more serious disruption than you had on this occasion. Consider if your entire server and workstation inventory required a total rebuild.....

If your issue has been solved, please use the "Mark as Solution" link on the most relevant thread.