Microsoft today updated KB 946676 which is the Knowledge Base article covering the Windows Home Server data corruption bug. This bug causes data to be corrupted when you try to edit or save certain files directly on the home server within certain applications.
As well as fixing the original publication date of the article, worrying, Microsoft states that customers have reported data corruption occurring with a growing list of applications. I guess they have a responsibility to report the full list of issues reported - however, according to the article Microsoft has only reproduced the bug on certain applications within that list.
In summary, Microsoft has reproduced the bug successfully with the following applications:
- Windows Vista Photo Gallery
- Windows Live Photo Gallery
- Microsoft Office OneNote 2007
- Microsoft Office OneNote 2003
- Microsoft Office Outlook 2007
- Microsoft Money 2007
- SyncToy 2.0 Beta
- Intuit QuickBooks
- uTorrent
Home Server users have reported the issue occurring with the following list of applications:
- Photoshop Elements
- Zune Software
- Apple iTunes
- TagScanner
- Mozilla Thunderbird
- Adobe Lightroom
- Intuit QuickenMS Digital Image Library
- MP3BookHelper
- ACDSee
- WinAmp
- Windows Media Player 11
- Microsoft Office Excel
- Visual DataFlex
So, to put this in perspective, the list of potential applications affected is growing. That is very worrying, particularly given the prevalence of some of the applications - iTunes, Excel, WMP11 etc.. Microsoft has not yet reproduced the issue on some of these applications but also they’ve not yet been discounted, so at any point, an application may move from the reported list to the proven list.
A few things to bear in mind:
1. Microsoft is actively working on the issue - as the KB article indicates, they have reproduced the issue and are working on a fix. Clearly that fix is not yet available, but I have been told it’s being dealt with as a very high priority. We (MVPs) are continually seeking updates from them on the issue.
2. Until the fix is issued, my advice is do not edit any file in any application directly on your home server. Copy the file to a home computer, edit it, and then copy it back again.
3. This issue occurs if a file is edited directly on the server when the server is under stress (ie. copying/moving a large amount of files). It’s difficult to replicate, but not impossible. I do not know how many have really been affected by the issue, but the potential is there. [Update: The notification regarding the server needing to be under stress appears to have been removed from the KB with this update. Sorry not to spot this!] The backup, remote access etc functions of your home server are not affected, so don’t think you have to pull the plug. Just be careful.
4. The issue only affects systems with more than one hard drive. If your home server has just one drive, you won’t be affected. If you’re running multiple drives, with folder duplication switched off, there is still a potential risk if you edit files directly on the server from a home computer.
We’re all hoping for a fix soon.
More Info: Microsoft Knowledge Base | Discuss Your Experiences
February 21st, 2008 at 3:01 pm
Thanks for the update on this issue! The fact that SyncToy 2.0 Beta is also affected is VERY frightening. I assume many WHS owners use SyncToy to copy/move their files to their Home Servers.
AndreasM
February 21st, 2008 at 3:02 pm
I cannot seem to reproduce the site where I found the information, but wasn’t this bug also determined to be linked to having multiple drives on the server? I may be wrong, since I can’t seem to find my source.
February 21st, 2008 at 3:04 pm
wow, I really should read the related links before posting (thought I read it on another blog). I withdraw the statement.
February 21st, 2008 at 6:21 pm
Hello Terry,
I’m afraid I can confirm the datacorruption for Excel files. In fact I had to take back a two disk WHS machine from a client because of it. As posted elsewhere on your Blog, I will extensively test the WHS with a single disk to see if that works out OK, so my client can have his server back. Furthermore I want to try out a Jstac Victoria unit, which is essentially a Raid1 controller, but not connected to the mainboard, so WHS won’t have to deal with drivers. My plan is to maintain a mirror of the system disk, without WHS even knowing the second disk is there.
I’m still glad that I supllied the WHS with 2 500Gb harddisks…
Keep up the good work, Wegotserved is one of the beste resources for WHS!
Mark Hazeleger
February 21st, 2008 at 6:37 pm
Thanks Mark,
Yes, we (WHS MVPs) were briefed by Microsoft that the data corruption bug only affects Windows Home Server systems with multiple hard drives. Even if you have folder duplication switched off on a multi-drive system, you are still at risk. We were briefed that single drive systems are safe.
I do not believe the issue is related to drivers, so I’m not sure whether the RAID controller workaround will be a solution, but do let us know how you go.
Glad you’re enjoying the site!
Terry
February 21st, 2008 at 7:06 pm
This is very disheartening. I have had to stop recommending the product as this bug completely invalidates using the server as a file share. I have had to move my active shares off the server onto a raid array and relegate my server to backup and long term storage.
It is not clear that they even have a solution under test yet.
Bummer…big bummer.
February 21st, 2008 at 7:21 pm
This is very disconcerting. Fortunately, I have a single drive system and I’ve resisted putting in a second drive because of this concern. But I feel really exposed with my data on a single drive (even though I have an online Idrive-E backup. Copying files to the local PC, editing them, then copying them back is a really awkward solution for a system that is supposed to protect your data.
February 21st, 2008 at 7:37 pm
Jim
Agreed - it’s not a solution, it’s a workaround and will keep your data safe.
Terry
February 21st, 2008 at 8:58 pm
I believe that I’ve had this same bug happen on my Home server three times. For a while I was using live photo gallery to edit pictures, and I kept getting warnings from the console after that. I would log into the console click the repair button and it would go away. Since they released the bug information I have stopped using live photo gallery, and I haven’t had the data corruption since. The interesting thing is that my Home Server has just one drive in it.
February 21st, 2008 at 9:09 pm
Hi Donnie
Your issue sounds unrelated to the data corruption bug - if you’d hit the bug, the images you had edited would be permanently corrupted. There’s no “repair” solution.
If you have one drive, you’re safe.
Terry
February 21st, 2008 at 9:32 pm
Well, the idea behind the Victoria unit is to provide redundancy without using unsupported technology withing the OS. Sure I have seen reports of people installing WHS on a regular (hardware) RAID array, but who knows what other quirks WHS may develop. Of course WHS is based on Windows Server 2003 which has no problems with RAID, but they have tinkered with it greatly, especially in the storage area, so I’m a bit wary introducing foreign bodies to WHS… WHS hopefully can’t wreck what it does not know about to begin with.
I’ll keep you posted on this workaround.
Mark
February 22nd, 2008 at 3:26 am
I can confirm the corruption issue with an Excel shared file and Quicbooks. That’s Excel 2003 and Quickbooks 2008 to be clear. I actually looked at the Quicbooks file in a hex editor after the corruption and compared it to a recent backup. It was like a warzone! Full blocks of data just cleared right out of the file, and large blocks and with no apparent pattern. I’m running 4 250gb drives on a PCI controller.
Running 2 big drives in a mirror seems like such a lame compromise. So is storing a shared file on a workstation and then letting WHS backup the workstation.
I plan to aim for 10tb or more with an 8 port sata controller. If MS can sort this out then the WHS folder duplication technology actually is superior to raid in disaster and recovery situations. With the promised external hard drive backup in power pack 1 we are talking about a nice package. HOWEVER, until this corruption issue is sorted out, this product is not ready for primetime.
February 22nd, 2008 at 3:29 pm
As this goes on, it makes me wonder if there is a fix in the works. I have this horrible feeling that they have discovered, or are discovering, that this is an inherent problem with the DE technology and it’s going to require a substantial rewrite of the product.
Microsoft needs to start being more open and let us know what they are doing to resolve the issue and give us an ETA on the fix.
February 22nd, 2008 at 4:06 pm
I ditto what Mike says above. I am really concerned that there is a design flaw. If MS has a fix in test they should say so NOW….but I am left with the feeling they are still in the characterization phase.
Terry, I take a big exception with your point 3. I am sure stress is a factor…but I played around putting my some of my OneNote notebooks on the server…and it corrupts very, very quickly wit simple updates to the notebooks even when there is no other activity on the server. In my opinion, people should not put any actively changed files on a WHS. Static files are fine…but if you are going to do anything to the file. Do it elsewhere.
I love WHS…..I love my HP box I am running it on….but this bug cripples the product. I think listing the apps with issues is a waste of time. I believe any application that changes a part of a file is affected by this bug. Only whole file operations are safe. MS should not be selling a server product that has data corruption bugs in the base file system.
February 22nd, 2008 at 4:25 pm
Hey Bob,
I think we’re aligned - point three came through to us from Microsoft as an influencing factor, so I’ve shared it (I haven’t experienced the bug myself). If you’ve had a different experience, then that sucks even more.
As stated above, our advice has always been not to directly edit any file on the home server in any application until the bug is fixed. With regard to where MS are in fixing the bug, and as to the underlying cause, I agree with you that we need to hear from MS directly as soon as possible - both I and others have made the point to them directly this week, don’t worry. Whether they act on that advice or not is entirely their perogative.
Best
Terry
February 22nd, 2008 at 6:18 pm
Bob
My bad - see update. Looks like the reference to the server needing to be under stress has been removed from the KB article. My apologies.
Terry
February 22nd, 2008 at 6:33 pm
I can personally agree that PST files will get corrupt. I was point folders to network locations and using offline files. Now i need to reteach the wife how to avoid this, and backup the whole server.
February 26th, 2008 at 4:41 pm
Well, I bought into this thing Christmas of 07, bought the two drive, 1TB HP MediaSmart. Wife isn’t very happy that she can’t do photo editing on the home server (how I sold her on it). It’s basically an $800 automated backup box for us. Well, that’s kind of ridiculous, but it’s also not the worst thing in the world. At least we’re finally backed up in some reasonable capacity.
As a former MS employee, I can say that this scenario is completely absurd. How something like this got through to RTM, may be unknowable.
However, MS owes early adopters of a seriously flawed v1.0 product something for this. This is “not a risk” of a v1.0 product–the corruption is a risk of a beta product.
MS should give all legitimate licensees of WHS 1.0 an free upgrade to the minor release at minimum, and if that was coming anyway, then a free upgrade to the Vail product.
February 26th, 2008 at 6:41 pm
Exactly Mike!
How could MS even let a product like this onto the market?!
Forget the number of apps effected, they should have at least tested Excel, one of their own Apps!
February 27th, 2008 at 10:38 am
I have been tring to download a free antivirus program from download.com or Cnet.com for about two months now,but everytime I suppose get it downloaded and go to install it, I get a message saying that it can’t be installed due to corupt files.Says try to download it again… But I don’t… I am currently running Windows Firewall and Windows Defender,and protection against mailware virus by what comes with PeoplePC.com…. Is it safe enough for me to continue doing World Wide Web Sereaches??? Please let me know… Thanks KYman41
February 29th, 2008 at 10:54 pm
Well I want to bitch and rage but how come any one that has used the beta never saw these problems? Maybe because the beta users were not normal users and were only using it for backup purpose, meaning they were “tech” from tech builders and were only experiencing the product in a backup oriented fashion object and never played with it like the normal consumer will and now that it is delivered to the common people that will use it in a normal way the bug just appeared.
Do you think that I am right on wrong on that beta phase pre release of Windows Home Server?
February 29th, 2008 at 11:06 pm
I think the fact that it wasn’t picked up in beta shows that it’s pretty rare bug - very serious, but fortunately, it’s not affected a huge amount of people.
Terry
March 1st, 2008 at 12:19 am
You are right and it won’t affect me as I am now aware of this.
Kudos for your site and Microsoft should has send me a warning note by the use of window update, if you can’t give and fix update at least tell the one that have the product on automatic update like a splash screen note to warn about and how to prevent this?
Good thing I have subscribed to your web site… good thing indeed.
Would that be a suggestion that you could carry in a wish list in the WHS Console to the WHS Team?
Michel
March 1st, 2008 at 6:20 am
I really home MS addresses this ASAP. I’m currently building my server atm and plan to use it as a central storage system for all the household PC’s. But the chance is always there a file will be edited on the server.. and this really scares the hell out of me. I’ve spent big $$ on this baby and I need stability MS! Hello! 160GB for the o/s and 4 - 1TB Drives in Stripe + Mirror. I’m thinking I’ll wait before I build the Arrays.
March 1st, 2008 at 7:38 am
If Michel
If you visit http://connect.microsoft.com/windowshomeserver you should be able to send in the request yourself!
Terry
March 4th, 2008 at 5:21 am
Hi Terry I seem to have all the time a problem to post to that site,
I have the “cookie” problems but will do it soon afer I decide to copy paste and do the process again clean up the IE options.
Thank you
March 5th, 2008 at 7:28 am
OK it’s now MARCH. MS has known about this since April, has been able to replicate it since December. Still no fix. HP has nothing to say. I bought my EX470 the week before christmas and immediatly added another 500 GB drive to take advantage of the replication features. I moved all my MP3’s over to it so I’d have a single repository, and within a fewweeks started noticing that a lot of files were suddenly not playing the last 3 - 5 seconds of a song. Fortunatly I kept my original MP3 directory on the other computer so I had a backup, but the idea was to rip everything to the EX470 directly going forward. I’m glad I didn’t move my original picture directories over (just the “album copies” I wanted to use via the media server function).
If something as simple as Windows Media Player 11 can corrupt data files I can’t trust anything to work with this box. I think a rebate of somekind is in order, but more importantly where the heck is the FIX?!?!
JT
March 7th, 2008 at 9:40 am
Worth adding that these problems DID come up during the Beta, although the reports were inconsistent. Obviously, now you look at the list of applications that are affected, different people were reporting what appeared to be different issues across a number of different usage scenarios. Where MS fell down was their inability to join the dots and appreciate that they had a SERIOUS problem on their hands. They should take a look at their analysis of the beta reports and review their practice, because the beta community were certainly reporting these issues during that time.
March 7th, 2008 at 7:26 pm
When using the WHS as a file server for Access data, any modification of the data via a frontend on a pc corrupts the file. Sometimes repairing the database with Access works, sometimes some records remain corrupted. So where do I store my address data now?