NT4 Server Faulty Hard Drive
Discussion
Hi Guys
I have an NT4 Server that I suspect has a faulty hard drive. Event log is showing "the driver detected a controller error on \device\harddisk0\partition1" at 2 minute intervals. Sometimes it's partition 2 instead of partition 1. These are my C: and D: drives.
What is the easiest and/or safest way to replace this faulty drive? I don't have any mirrors, stripe sets, etc. Just a plain 18GB SCSI drive with two primary partitions. There is an HP Surestore DAT40 SCSI drive attached and working if that helps.
Cheers
Phil
I have an NT4 Server that I suspect has a faulty hard drive. Event log is showing "the driver detected a controller error on \device\harddisk0\partition1" at 2 minute intervals. Sometimes it's partition 2 instead of partition 1. These are my C: and D: drives.
What is the easiest and/or safest way to replace this faulty drive? I don't have any mirrors, stripe sets, etc. Just a plain 18GB SCSI drive with two primary partitions. There is an HP Surestore DAT40 SCSI drive attached and working if that helps.
Cheers
Phil
The most conventional method is backing it up, replacing the drive then restoring it.
Whether I'd do that or not depends on your backup software, how difficult a bare metal restore is, and how complex the config of the machine is.
Failing that you could add another SCSI drive and setup software mirroring, then break the mirror afterwards (paying careful attention to the boot.ini file).
D
Whether I'd do that or not depends on your backup software, how difficult a bare metal restore is, and how complex the config of the machine is.
Failing that you could add another SCSI drive and setup software mirroring, then break the mirror afterwards (paying careful attention to the boot.ini file).
D
_DeeJay_ said:
The most conventional method is backing it up, replacing the drive then restoring it.
Whether I'd do that or not depends on your backup software, how difficult a bare metal restore is, and how complex the config of the machine is.
Failing that you could add another SCSI drive and setup software mirroring, then break the mirror afterwards (paying careful attention to the boot.ini file).
D
Whether I'd do that or not depends on your backup software, how difficult a bare metal restore is, and how complex the config of the machine is.
Failing that you could add another SCSI drive and setup software mirroring, then break the mirror afterwards (paying careful attention to the boot.ini file).
D
Hi DeeJay
The server is running Exchange 5.5, and I use NTBackup to do online backup every night.
I like the sound of mirroring; how error prone is it? And if it all goes wrong, is there a fall back? Also, what kind of downtime are we talking about?
Cheers
Phil
pcwilson said:
_DeeJay_ said:
The most conventional method is backing it up, replacing the drive then restoring it.
Whether I'd do that or not depends on your backup software, how difficult a bare metal restore is, and how complex the config of the machine is.
Failing that you could add another SCSI drive and setup software mirroring, then break the mirror afterwards (paying careful attention to the boot.ini file).
D
Whether I'd do that or not depends on your backup software, how difficult a bare metal restore is, and how complex the config of the machine is.
Failing that you could add another SCSI drive and setup software mirroring, then break the mirror afterwards (paying careful attention to the boot.ini file).
D
Hi DeeJay
The server is running Exchange 5.5, and I use NTBackup to do online backup every night.
I like the sound of mirroring; how error prone is it? And if it all goes wrong, is there a fall back? Also, what kind of downtime are we talking about?
Cheers
Phil
Yuk - Exchange 5.5. is a pain, and is nolonger supported (as NT4 isn't).
Downtime - the amount of time it takes to power the server down, install the disk and power it back up again.
You can then mirror the disks which will have a performance impact, but won't prevent access to the server.
Once finished, it's more downtime as you change boot.ini, then remove the old disk (and break the mirror)
It's been a while since I've done any of this on NT4 so I'll have to check my facts about software mirroring, though I'm 90% sure I'm correct.
Will it go wrong? Well, I've never seen it fail. I suppose the only risk is that the OS will try to copy a corrupt part of the disk and upset it, but then again, so would a backup.
The mirror process is not destructive (it literally just block copies the disk) so i'd say it was low risk.
d
pcwilson said:
I like the sound of mirroring; how error prone is it? And if it all goes wrong, is there a fall back? Also, what kind of downtime are we talking about?
Software mirroring (also known as RAID1) whilst not as efficient or reliable as hardware mirroring using a dedicated card is far more reliable than simply relying a single Hard Disk Drive (HDD).
I would recommend RAID1 for a basic setup with a daily backup to another medium, ideally something you take off-site. I'm not sure what you mean by 'online' backup with NTBackup - do you backup off-site automatically already using that (I didn't realise it could) or do you mean you backup whilst the server is running? If you're not backing up off-site, the simplest and cheapest solution is to use NTBackup to backup to an External HDD (EHDD - available for less than £80) and take that off-site with you. If you want something more sophisticated such as full disk image backup with disaster recovery to an EHDD or automatic off-site backup over your Internet connection then there are plenty of choices out there and my company specialises in these. I believe pastic_pig also does off-site data backup.
HTH
roadie.
I'm sorry, but that's just not good advice. So, what's your solution? Hardware RAID? You want to put a RAID controller into a really old server running unsupported software then install the drivers on the unsupported OS (which probably don't exist).
Assuming all that works, then what?
Not a good plan. If what he wants is to replace the existing hard disk drive with another one, relatively safely, I'd go with my orignal plan. If he wants to increase the fault tolerance of the machine, I'd suggest building a new machine altogether, in the same Exchange organisation and moving the mailboxes, perhaps even upgrade to Exchange 2003 at the same time. However, that is a lot of work and by the time he's done it, the disk may well have died.
You stated software RAID was unreliable. Can you recount a time when it has failed? Years ago, I used it a lot and never once did I have a problem with NT software disk mirroring. It's a means to an end (i.e. a way of copying the data whilst the machine is up), not a long term solution, and I seriously doubt it will fail.
Assuming all that works, then what?
Not a good plan. If what he wants is to replace the existing hard disk drive with another one, relatively safely, I'd go with my orignal plan. If he wants to increase the fault tolerance of the machine, I'd suggest building a new machine altogether, in the same Exchange organisation and moving the mailboxes, perhaps even upgrade to Exchange 2003 at the same time. However, that is a lot of work and by the time he's done it, the disk may well have died.
You stated software RAID was unreliable. Can you recount a time when it has failed? Years ago, I used it a lot and never once did I have a problem with NT software disk mirroring. It's a means to an end (i.e. a way of copying the data whilst the machine is up), not a long term solution, and I seriously doubt it will fail.
Edited by _DeeJay_ on Tuesday 5th September 15:32
Well, I'm setting up a test server running NT4 Server SP6a with two 9GB SCSI hard drives, and I'm going to practice creating / breaking mirrors etc...
One question, since I'm in the fortunate position of knowing my HDD is going to fail, once mirroring has completed successfully, do I break the mirror set before removing the faulty HDD, or do I remove the faulty HDD then break the mirror?
Also, what changes do I need to make to boot.ini, and when do I make them?
I've taken a brick level backup (using exmerge) of the Exchange mailboxes to a different physical drive in the same server, and I've moved the users data share (.doc, .xls etc...) to this other drive also.
My other plan was to rebuild my test server as a production server (BDC) and install exchange as new server in same site, then use move mailbox wizard, but the mirroring looks favourite at the moment.
I'm just trying to nurse this old server through for a few more months while we plan the upgrade to W2K3 and EX2K3.
One question, since I'm in the fortunate position of knowing my HDD is going to fail, once mirroring has completed successfully, do I break the mirror set before removing the faulty HDD, or do I remove the faulty HDD then break the mirror?
Also, what changes do I need to make to boot.ini, and when do I make them?
I've taken a brick level backup (using exmerge) of the Exchange mailboxes to a different physical drive in the same server, and I've moved the users data share (.doc, .xls etc...) to this other drive also.
My other plan was to rebuild my test server as a production server (BDC) and install exchange as new server in same site, then use move mailbox wizard, but the mirroring looks favourite at the moment.
I'm just trying to nurse this old server through for a few more months while we plan the upgrade to W2K3 and EX2K3.
Edited by pcwilson on Tuesday 5th September 15:54
pcwilson said:
Well, I'm setting up a test server running NT4 Server SP6a with two 9GB SCSI hard drives, and I'm going to practice creating / breaking mirrors etc...
One question, since I'm in the fortunate position of knowing my HDD is going to fail, once mirroring has completed successfully, do I break the mirror set before removing the faulty HDD, or do I remove the faulty HDD then break the mirror?
Also, what changes do I need to make to boot.ini, and when do I make them?
One question, since I'm in the fortunate position of knowing my HDD is going to fail, once mirroring has completed successfully, do I break the mirror set before removing the faulty HDD, or do I remove the faulty HDD then break the mirror?
Also, what changes do I need to make to boot.ini, and when do I make them?
Personally, I'd:
1) Create a boot disk which will boot either disk (i.e. the format a disk and copy the boot files to it, modifying boot.ini on each)
2) Ensure you've got a current backup on tape (using an Exchange aware backup utility)
3) Modify boot.ini on the server to point to the new disk partition. I assume you're installing a second disk of the same time and keeping the same partitions (including any HP system partitions etc). This should just be a case of changing rdisk parameters - see here www.microsoft.com/technet/prodtechnol/windows2000serv/reskit/prork/prbd_std_ccef.mspx?mfr=true)
4) Try to boot on the new disk. If it boots, remove the mirror, if it doesn't use your boot disk to boot the old one.
D
DeeJay you just didn't read my post properly. I have no idea why the tone of your post is as it is.
No actually, I wasn't suggesting he use hardware RAID, I was answering his question, "I like the sound of mirroring; how error prone is it? And if it all goes wrong, is there a fall back?" I'm well aware of the issues with RAID drivers and older Operating Systems.
No, I didn't say that at all.
I stated it wasn't as reliable as hardware RAID and I stand by that. I have never personally had a problem with software RAID but the fact is that hardware RAID is the superior solution for reasons which I'm sure are obvious, part from outright performance. However, I'm well aware that isn't the appropriate solution here, I was simply trying to supply information about the subject.
So, in summary, I suggest you wind your neck and in read the posts of others properly before replying so aggresively to someone who quite probably knows every bit as much as you about this and was simply trying to help another PHer. Your post has made me extremely angry and it's been a long time since anyone on here as managed that.
_DeeJay_ said:
I'm sorry, but that's just not good advice. So, what's your solution? Hardware RAID? You want to put a RAID controller into a really old server running unsupported software then install the drivers on the unsupported OS (which probably don't exist).
No actually, I wasn't suggesting he use hardware RAID, I was answering his question, "I like the sound of mirroring; how error prone is it? And if it all goes wrong, is there a fall back?" I'm well aware of the issues with RAID drivers and older Operating Systems.
_DeeJay_ said:
You stated software RAID was unreliable. Can you recount a time when it has failed? Years ago, I used it a lot and never once did I have a problem with NT software disk mirroring. It's a means to an end (i.e. a way of copying the data whilst the machine is up), not a long term solution, and I seriously doubt it will fail.
No, I didn't say that at all.
I stated it wasn't as reliable as hardware RAID and I stand by that. I have never personally had a problem with software RAID but the fact is that hardware RAID is the superior solution for reasons which I'm sure are obvious, part from outright performance. However, I'm well aware that isn't the appropriate solution here, I was simply trying to supply information about the subject. So, in summary, I suggest you wind your neck and in read the posts of others properly before replying so aggresively to someone who quite probably knows every bit as much as you about this and was simply trying to help another PHer. Your post has made me extremely angry and it's been a long time since anyone on here as managed that.
Sorry - I didn't mean to annoy, just obviously did not understand you post.
It came across as preaching, and superfluous to this particular issue. I'm currently at home when I should be at work (doing work area recovery as it happens) and drugged up to the eye balls, so I should perhaps stop posting until I'm better so I don't take my frustration out on anyone else!
For the record (and I hope this does not annoy), an 'online backup' in that context means backing up Exchange 5.5 whilst it is online (i.e. the database is running).
It came across as preaching, and superfluous to this particular issue. I'm currently at home when I should be at work (doing work area recovery as it happens) and drugged up to the eye balls, so I should perhaps stop posting until I'm better so I don't take my frustration out on anyone else!
For the record (and I hope this does not annoy), an 'online backup' in that context means backing up Exchange 5.5 whilst it is online (i.e. the database is running).
_DeeJay_ said:
Sorry - I didn't mean to annoy, just obviously did not understand you post. It came across as preaching, and superfluous to this particular issue.
I don't know about preaching, I certainly didn't write or read it that way. However, the reason I didn't give specifics was because I didn't feel I needed to add anything to what you had already suggested, other than some general information around the subject. If someone is using a single disk without RAID on a server then suggesting RAID and backup strategies is hardly unwarranted.
_DeeJay_ said:
I'm currently at home when I should be at work (doing work area recovery as it happens) and drugged up to the eye balls, so I should perhaps stop posting until I'm better so I don't take my frustration out on anyone else!
That doesn't sounds like fun!
_DeeJay_ said:
For the record (and I hope this does not annoy), an 'online backup' in that context means backing up Exchange 5.5 whilst it is online (i.e. the database is running).
No that doesn't annoy.
I suspected that as I hinted at when I said I didn't think NTBackup could do that (although of course it can if an off-site server can be made to appear local to the OS which is certainly achievable), but I've found that people mix the terms up - to some people "online backup" means backing up over an Internet connection, to others it means backing up whilst the 'service' is live/running/online. Anyway, no harm done. Good luck with the recovery work.
_DeeJay_ said:
Personally, I'd:
1) Create a boot disk which will boot either disk (i.e. the format a disk and copy the boot files to it, modifying boot.ini on each)
2) Ensure you've got a current backup on tape (using an Exchange aware backup utility)
3) Modify boot.ini on the server to point to the new disk partition. I assume you're installing a second disk of the same time and keeping the same partitions (including any HP system partitions etc). This should just be a case of changing rdisk parameters - see here www.microsoft.com/technet/prodtechnol/windows2000serv/reskit/prork/prbd_std_ccef.mspx?mfr=true)
4) Try to boot on the new disk. If it boots, remove the mirror, if it doesn't use your boot disk to boot the old one.
D
Hi DeeJay,
Sorry for being a total muppet, but:
1) Other than boot.ini, what other files do I copy to the boot floppy?
2) Check.
3) My faulty disk is 18GB (6GB C: + 12GB D: ). I've moved everything off D: so it's now redundant. I only have a 9GB disk to hand which I'm going to install and mirror this with C:. Should this work?
4) I may need to come back to you when I get his far, if that's okay?
I really do appreciate all the advice; PH truely is the font of all knowledge.
Would you believe, in nearly 8 years, this is the first time I've had to replace a failing hard disk? (quickly touches something wooden! oh ah misses) I back up everything to tape every night (using exchange aware nt backup) and I've only ever had to restore the odd word or excel file that someone has deleted accidently. Now that I've built a test server, I really need to test my Exchange disaster recovery plan.
P.S. My mum always said I could start a fight in an empty room
_Deejay_ said:
I've got to nip to the doctors - will be back in 30 mins and will answer properly.
I think NT 4.0 mirrored partitions rather than disks, though I could be wrong. Something to check whilst I'm away
D
I think NT 4.0 mirrored partitions rather than disks, though I could be wrong. Something to check whilst I'm away
D
Yep. NT4 disk administrator lets me mirror partitions rather than disks. Having said that, I haven't successfully managed this on my test server because it keeps blue screening on me
NT mirrors partitions not disks so you should get away with the smaller disk as long as everything really is off D:
files needed for a boot floppy are (on an intel box)
msdos.sys
io.sys
ntdlr
ntdetect.com
boot.ini
and if your boot.ini has "scsi" not "multi" when describing the disk layout then you will need ntbootdd.sys (which is just the aspi driver for your scsi card copied to that name)
once you have setup you boot.ini and the mirror etc on your test server you might find that if you boot from the second disk while the first disk is still available it will give you a stop7b inaccessable boot device error. If you unplug the power to the first disk then redo the boot it should come up ok.
On thing to be aware of is if your disk is failing when you come to do this for real on the live server you are going to need to shut it down. Keep the dowtime to the absolute minimum as failing disks really don't like to cooldown if they have been spinning for X years.
You might want to do and offline backup of exchange and OS just before you go for the new drive install. ( can't have too many backups )
Good luck
files needed for a boot floppy are (on an intel box)
msdos.sys
io.sys
ntdlr
ntdetect.com
boot.ini
and if your boot.ini has "scsi" not "multi" when describing the disk layout then you will need ntbootdd.sys (which is just the aspi driver for your scsi card copied to that name)
once you have setup you boot.ini and the mirror etc on your test server you might find that if you boot from the second disk while the first disk is still available it will give you a stop7b inaccessable boot device error. If you unplug the power to the first disk then redo the boot it should come up ok.
On thing to be aware of is if your disk is failing when you come to do this for real on the live server you are going to need to shut it down. Keep the dowtime to the absolute minimum as failing disks really don't like to cooldown if they have been spinning for X years.
You might want to do and offline backup of exchange and OS just before you go for the new drive install. ( can't have too many backups
) Good luck
Article on boot disks can be found here: www.windowsnetworking.com/kbase/WindowsTips/WindowsNT/AdminTips/Booting/CreateaWindowsNTbootdisk.html though Malman's pretty much covered it.
D
D
Hi DeeJay
Apologies for the delay in responding. Just got my kids back from Girls Brigade and into their beds.
I'll make my boot disks in the morning and give it a try. Thanks again for all the help; it really is much appreciated.
One (hopefully) last question. How do you do an offline backup of the OS? I backup the entire C: drive to tape every night using NTBackup but the backup log is full of errors relating to files that couldn't be backed up e.g dhcp, dns etc... and I've never really trusted that I'd be able to do much with this should the need arise. Whenever I've had OS corruption before, I've always done a parallel install or a fresh install, and then reinstalled all apps.
Cheers again
Phil
Apologies for the delay in responding. Just got my kids back from Girls Brigade and into their beds.
I'll make my boot disks in the morning and give it a try. Thanks again for all the help; it really is much appreciated.
One (hopefully) last question. How do you do an offline backup of the OS? I backup the entire C: drive to tape every night using NTBackup but the backup log is full of errors relating to files that couldn't be backed up e.g dhcp, dns etc... and I've never really trusted that I'd be able to do much with this should the need arise. Whenever I've had OS corruption before, I've always done a parallel install or a fresh install, and then reinstalled all apps.
Cheers again
Phil
I'm not sure to be honest - it's been a long time since I touched Windows NT 4.0. When I did backup NT machines I used a third party backup product (usually ArcServe) with an open file agent to ensure all files were backed up.
From what I remember NTBackup on NT 4.0 didn't have the ability to backup open files. You can backup DNS, WIN, DHCP etc individually prior to the OS backup using other tools, or if you're able, stop the services prior to the final backup.
Microsoft admit this is a failing in NT 4 NTBackup here - www.microsoft.com/technet/archive/winntas/tips/techrep/backup.mspx?mfr=true
They also pretty much suggest a third party backup product.
Sorry I couldn't be of any more help. However, to be honest the most important part of the server sounds to be the Exchange online/offline backup. You should be able to rebuild a machine with the same name, install Exchange and restore the files to recover Exchange if it all goes Pete Tong.
Darren.
From what I remember NTBackup on NT 4.0 didn't have the ability to backup open files. You can backup DNS, WIN, DHCP etc individually prior to the OS backup using other tools, or if you're able, stop the services prior to the final backup.
Microsoft admit this is a failing in NT 4 NTBackup here - www.microsoft.com/technet/archive/winntas/tips/techrep/backup.mspx?mfr=true
They also pretty much suggest a third party backup product.
Sorry I couldn't be of any more help. However, to be honest the most important part of the server sounds to be the Exchange online/offline backup. You should be able to rebuild a machine with the same name, install Exchange and restore the files to recover Exchange if it all goes Pete Tong.
Darren.
Gassing Station | Computers, Gadgets & Stuff | Top of Page | What's New | My Stuff