Solved

How to Tell if a Server is RAID 5 enabled...

Posted on 2001-08-14
19
222 Views
Last Modified: 2013-11-14
I have two dedicated web servers that are managed by a hosting company.  We were guaranteed that the servers would be configured for RAID 5 and fully clustered for fail over purposes.  Not that I don't trust them, but I want to ensure that the systems are working properly.  I would like to find out how a RAID 5 clustered system should "look" and if my system is not correct, how should I explain to them what needs fixed.  I would like make this worth a good amount of points in hopes of a detailed explanation, but any assistance is appreciated.

I have root access and I know that each system has the following configuration:
- Dual PIII 1Ghz Processors
- Three 18Gb SCSI UltraWide Drives
- Both show a RAID Controller ERROR under Other Devices but have a QLOGIC QLA12160, 64 BIT Dual PCI 160M SCSI HBA controller installed under SCSI & RAID devices.

One system is used for a web server and the other for SQL server, but SQL server is only installed on one machine.  If the servers are supposed to be clustered, shouldn't SQL server be installed on both machines?

For one system the disk manager shows:
   Disk 0 - Drive C: - Type=Basic   - Layout=Partition
   Disk 1 - Drive E: - Type=Dynamic - Layout=Simple
   DIsk 2 - Drive F: - Type=Dynamic - Layout=Simple

For the other system, the disk manager shows:
   Disk 0 - Drive C: - Type=Dynamic - Layout=Mirror
   Disk 1 - Drive C: - Type=Dynamic - Layout=Mirror
   Disk 2 - Drive E: - Type=Dynamic - Layout=Simple

I was told that one complete system could go down along with two of the hard drives from the other system and my sites would still stay up. Does this look correct for that type of configuration?  
0
Comment
Question by:bmccleary
  • 11
  • 5
  • 2
  • +1
19 Comments
 
LVL 7

Expert Comment

by:pbessman
ID: 6386533
Seeing as the disk manager is indicating mirror I would assume that the MIRROR is working correctly.  What you are seeing as an error may be some other RAID device on their system.  Perhaps their system includes a RAID controller that only supports Level "0".  Since your needs are for RAID "1" or higher I am assuming they have disabled the card with the error by not loading the drivers for it and have enabled the add on card by configuring its drivers.  FOr more info on the levels of RAID, check here:  http://www.vogon-data-recovery.com/disk_recovery-04.htm
0
 
LVL 13

Expert Comment

by:magarity
ID: 6386998
Just add up the capacity!

A RAID level 5 is:
[ (# of drives) times (smallest drive's capacity) ] minus (smallest drive's capacity)
Thus: A RAID 5 consisting of three 20GB drives would weigh in at 40GB.  

Of course, so would a RAID 0 with one drive disabled, so this isn't a perfect test... but it will give you a start!

regards,
magarity
0
 
LVL 7

Expert Comment

by:pbessman
ID: 6387003
RAID is on remote server not home machine???
0
 
LVL 13

Expert Comment

by:magarity
ID: 6387005
Whups, I just noticed that you do list your drives' capacities: 18GB.  So a RAID 5 of 3 drives of 18GB capacity should result in a total usable space of 36GB.  If it's 54 then you're set up with a RAID 0.

The real trick is whether the stripe size was optimized to your application, but that's another story and only really has bearing if your server's load is fairly high.  The SQL server should probably have a smaller stripe size than the web server.  Aaaaa, esoteria alert!

regards,
magarity
0
 
LVL 13

Expert Comment

by:magarity
ID: 6387018
I keep posting because I really like this question; that setup looks suspicious.  BUT:

If you could please supply a little more information:
1.  What are the reported capacities versus the 18GB actual?
2. Is that disk manager from Windows2000 Server/Advanced Server?  I have used several RAIDs of different types with these operating systems, but I need to know for certain that's what you're using.  If not, what operating system is it?

Assuming that this IS Windows2000 Server/Adv Server, then that disc manager report indicates these people are using the software RAID feature.  It does NOT do RAID 5.  It does RAID 1 and 0.  If you only have 3 drives, it really looks like the mirror is across partitions, not drives.  A physical failure of a complete disk would leave you SOL in that case.

Meanwhile, the real problem is that the QLogic QLA12160 is not a RAID controller.  Windows 2000 Server will do a software RAID 1 and 0 , but that means every disk operation takes a lot of the CPU's attention.  Your machine's performance is going to be hating it as soon as you get a decent load on it.

regards,
magarity
0
 
LVL 13

Expert Comment

by:magarity
ID: 6387026
One last post, then I have to get some sleep.

"if my system is not correct, how should I explain to them what needs fixed."

This is the easiest part!  If your setup is what I think it is (we'll need to do some more back-and-forth before deciding) then the way to explain how to fix it is to call some other hosting service and move your website there instead.  If you've been told you have a RAID 5 on different disks and really have a software mirror on multiple partitions, then that would be a case of unforgivable gross incompetence.

regards,
magarity
0
 
LVL 13

Expert Comment

by:magarity
ID: 6387028
BTW, I see from your profile that you're in Denver.  Hi, I'm in Fort Collins!
0
 
LVL 4

Expert Comment

by:1175089
ID: 6387124
Hi, I am from Lovech, Bulgaria, but your partition sheme is suspicious:
If you has hardware RAID5 for example you cant see mirror partitions on Disk Administrator.You can see mirror partitions only if you use software mirror (RAID 1)from the OS.The real structure of the RAID you can see only with special software that usually comes with RAID adapter (For example I use ASUSTek GUI RAID Manager for my DA2100 raid controller)
0
 
LVL 13

Expert Comment

by:magarity
ID: 6388060
Now that it's morning and I've had some sleep, it does look like for catastrophic failure, drive 'C' of the second machine is covered:

Disk 0 - Drive C: - Type=Dynamic - Layout=Mirror
Disk 1 - Drive C: - Type=Dynamic - Layout=Mirror

Drive 'C' is clearly indicated as a mirror over two physical disks.  If one of them fails, there will be no interruption of service.  All other disks still appear to be on individual disks...  as far as I can tell, the QLogic QLA12160 is not a RAID controller, but is often used to control self contained RAID devices because of its very high throughput.  Very mysterious...

I really need to know:
On the first machine, are drives C, E, and F all 18GB?
On the second machine, are drives 'C' and 'E' both 18GB?
With Win2k, use the regular 'my computer' route to check this, rather than the disk manager.

regards,
magarity
                                         
0
6 Surprising Benefits of Threat Intelligence

All sorts of threat intelligence is available on the web. Intelligence you can learn from, and use to anticipate and prepare for future attacks.

 

Author Comment

by:bmccleary
ID: 6388723
Whew!  What great responses!  Thanks to all.

pbessman - Thanks for the link. And I am not sure which machine, if any, is set up correctly.

1175089 - If you can only truly see a RAID system through the hardware drivers, it is possible that I do have it set up?

-------------------------------------------------

Magarity - Thanks for all the input.  Here is the information that you requested.
-Both boxes are running Win2000 Advanced Server

Through My Computer:
- The first machine (the one with C, E & F Drives), all drives are showing a 17GB total capacity (which is strange because they are 18GB drives)
- The second machine (the one with a C mirrored & E drive), both drives are showing 17GB

- None of the drives are showing being split into multiple partitions with the exception on the second machine, there is a small 8MB partition on the C: mirror drives, that I was told was for a "shock absorber" in case the disks filled up.

- BTW - does SQL server need to be installed on both machines to cover incase of fail over, or does this configuation not even look like it is clustered?

You talk about the QLogic card, but what is a "self contained RAID device".  

Also, you had mentioned switching hosting companies, but unfortunately, that is not an option.  I just spent 2 months moving 50 clients over to these new boxes and setting them up on email - I'm not doing that again.  These guys are pretty good with the managed services and I do have full control over the boxes.  But if you feel up to it, the boxes are actually located in Northglenn and if you would like to have a quick look on a contractual basis and offer any ideas, perhaps that could be in order.
0
 

Author Comment

by:bmccleary
ID: 6388791
If it helps, here are screen shots of the configuration:
http://206.61.99.201/sys1_config.gif
http://206.61.99.201/sys2_config.gif
0
 
LVL 13

Expert Comment

by:magarity
ID: 6388940
The screenshots were perfect...  The only RAID on these systems is a software mirror RAID on system #2, comprising two of the disks.

What this means for calamity coverage:  If machine #1's drive 0 or drive 1 has a total physical failure, operation will remain uninterrupted.  ANY OTHER FAILURE of the physical disks is NOT covered and will result in failure of any functionality requiring those disks.  This includes all disks of machine #1.

"17GB total capacity (which is strange because they are 18GB drives)"  
Raw drive capacities vary from formatted capacity mainly because depending on whether you format with FAT32, NTFS, EX2FS, or whatever, you get different usable amounts.  NTFS, for example,  uses more space than FAT32 for formatting information, but offers better performance for finding files.  The trade-off is well worth it.

"You talk about the QLogic card, but what is a "self contained RAID device". "
This is an external device full of hard drives that does its own RAID logic internally.  As far as the SCSI controller is concerned, it is one simple SCSI drive.  This is not what you have unless you paid a LOT extra for it.  And it wouldn't show up as three drives in the disk manager.

If you have full control of the boxes, this is the solution:
1. Find out what the details on the RAID controller listed as " RAID Controller ERROR under Other Devices".  You have paid for a Dell with a RAID controller?  You will need to determine its status.  Is it defective?  Does it still exist in the system or has it been removed?  If not removed, why is it not enabled?  Why is the QLogic card being used in its stead?

2.  Back up EVERYTHING, because converting between RAID levels can only be done with a complete system re-installation.  This can't be attempted before the questions in part 1 have been satisfied.  Unfortunately, without the RAID card properly configured, your installation of Win2000 expects to boot from a QLogic card, and not a whatever the RAID controller really is...  This leads to some serious issues, and it is doubful that a simple backup-current-install then restore-to-RAID would boot properly.  Just to be safe, I would strongly suggest that the machines be reinstalled once the RAID controller is worked out.

3.  These machines may or may not be configured to fail to each other, but it is unapparent from looking at the disk manager.  I would suggest waiting for a slow time for your website and remote-rebooting one of them.  If service continues uninterrupted for the three minutes it takes to reboot one, then obviously that has been setup correctly.  In addition to being a simple test you can do yourself, when it fails (I'm only partly certain this will happen, but given the configuration of these disks, I wouldn't take it for granted that these people have set anything up correctly) you can call the hosting people and voice your concerns about this setup.

4.  Windows 2000 Advanced Server is the proper version to do a redundant cluster (regular Server cannot), so you're all set there.  I don't have the networking information on these two systems, and while I've used 2kAS, I've never set up a cluster (only read the manual on how), so I wouldn't be an apropriate choice to officially consult on this.  Thanks for the offer, though!

***********
Additional considerations:

1.  While you've obviously read that having two redundant machines is better than one.  However, you have both of them at the same location.  If the power fails at this site, their T1 or whatever goes down, then your fancy fail-over system is worthless.  Ideally, use two locations, and ideally one on each coast.  If nothing else, at least have them on different sides of the city.

2.  Your idea of fail-over may be different from mine.  Does each machine have its own independent 'regular job', so to speak, and is just supposed to be able to take over for the other in emergencies?  Or, are both machines supposed to be doing the same job all the time and just under an extra heavy load if one goes down?  If it is a case of independent regular jobs, then point 1 above is less important.

regards,
magarity
0
 
LVL 13

Expert Comment

by:magarity
ID: 6388987
The screenshots were perfect...  The only RAID on these systems is a software mirror RAID on system #2, comprising two of the disks.

What this means for calamity coverage:  If machine #1's drive 0 or drive 1 has a total physical failure, operation will remain uninterrupted.  ANY OTHER FAILURE of the physical disks is NOT covered and will result in failure of any functionality requiring those disks.  This includes all disks of machine #1.

"17GB total capacity (which is strange because they are 18GB drives)"  
Raw drive capacities vary from formatted capacity mainly because depending on whether you format with FAT32, NTFS, EX2FS, or whatever, you get different usable amounts.  NTFS, for example,  uses more space than FAT32 for formatting information, but offers better performance for finding files.  The trade-off is well worth it.

"You talk about the QLogic card, but what is a "self contained RAID device". "
This is an external device full of hard drives that does its own RAID logic internally.  As far as the SCSI controller is concerned, it is one simple SCSI drive.  This is not what you have unless you paid a LOT extra for it.  And it wouldn't show up as three drives in the disk manager.

If you have full control of the boxes, this is the solution:
1. Find out what the details on the RAID controller listed as " RAID Controller ERROR under Other Devices".  You have paid for a Dell with a RAID controller?  You will need to determine its status.  Is it defective?  Does it still exist in the system or has it been removed?  If not removed, why is it not enabled?  Why is the QLogic card being used in its stead?

2.  Back up EVERYTHING, because converting between RAID levels can only be done with a complete system re-installation.  This can't be attempted before the questions in part 1 have been satisfied.  Unfortunately, without the RAID card properly configured, your installation of Win2000 expects to boot from a QLogic card, and not a whatever the RAID controller really is...  This leads to some serious issues, and it is doubful that a simple backup-current-install then restore-to-RAID would boot properly.  Just to be safe, I would strongly suggest that the machines be reinstalled once the RAID controller is worked out.

3.  These machines may or may not be configured to fail to each other, but it is unapparent from looking at the disk manager.  I would suggest waiting for a slow time for your website and remote-rebooting one of them.  If service continues uninterrupted for the three minutes it takes to reboot one, then obviously that has been setup correctly.  In addition to being a simple test you can do yourself, when it fails (I'm only partly certain this will happen, but given the configuration of these disks, I wouldn't take it for granted that these people have set anything up correctly) you can call the hosting people and voice your concerns about this setup.

4.  Windows 2000 Advanced Server is the proper version to do a redundant cluster (regular Server cannot), so you're all set there.  I don't have the networking information on these two systems, and while I've used 2kAS, I've never set up a cluster (only read the manual on how), so I wouldn't be an apropriate choice to officially consult on this.  Thanks for the offer, though!

***********
Additional considerations:

1.  While you've obviously read that having two redundant machines is better than one.  However, you have both of them at the same location.  If the power fails at this site, their T1 or whatever goes down, then your fancy fail-over system is worthless.  Ideally, use two locations, and ideally one on each coast.  If nothing else, at least have them on different sides of the city.

2.  Your idea of fail-over may be different from mine.  Does each machine have its own independent 'regular job', so to speak, and is just supposed to be able to take over for the other in emergencies?  Or, are both machines supposed to be doing the same job all the time and just under an extra heavy load if one goes down?  If it is a case of independent regular jobs, then point 1 above is less important.

regards,
magarity
0
 
LVL 13

Expert Comment

by:magarity
ID: 6388993
Sorry about the duplicate; accidentally hit 'reload'...

Anyway - I've been telling you what it shouldn't look like and forgot to tell what it should like like!  Instead of:

Disk 0 - Drive C: - Type=Basic   - Layout=Partition Size = 17GB
Disk 1 - Drive E: - Type=Dynamic - Layout=Simple Size = 17GB
DIsk 2 - Drive F: - Type=Dynamic - Layout=Simple Size = 17GB

A properly configured RAID 5 on this system will look like this:

Disk 0 - Drive C: - Type=Basic - Layout= Partition  Size = 34GB

regards,
magarity
0
 

Author Comment

by:bmccleary
ID: 6389109
Magarity,
The Other Devices - RAID Controller error doesn't have any information listed for manuafacturer or device type, but it does have a location, which it means that it is installed.

I am assuming that you mean machine #2 not #1 can have a drive 0 or 1 failure.

Both machines have their independant jobs (one for web services, the other for database services) and they are meant to take over the other job in the case of a failure.

For the drive configuration, it wouldn't show three seperate drives with only one logical drive id?

If I can prove that the system is not setup correctly, I can have them set me up two new servers in the proper confing and then move my data over.  I just need to make sure that I am sounding intelligent about how it needs set up, and not like a complaining a-hole.  Do you have any good references that I can go to for an overview on the proper configuration of a clustered server?

Thanks!
0
 
LVL 13

Accepted Solution

by:
magarity earned 250 total points
ID: 6389231
1.  "but it does have a location, which it means that it is installed."

Then you will need to ask why it is not being used and why the QLogics have been substituted.  Are these their machines or yours?  If they belong to the hosting company, all you can do is complain that you were promised RAID 5 on all disks and are getting only RAID 1 on two of them.  If these are your machines, you should demand to know who has been monkeying with your hardware and why.  With such an obscure error message, you'll need to check the invoice for the machine to determine its make and model of RAID controller or actually go look in the machine and see for yourself.

2.  "assuming that you mean machine #2 not #1 can have a drive 0 or 1 failure."

Yes, typo, sorry.

3.  "For the drive configuration, it wouldn't show three seperate drives with only one logical drive id?"

No.  That's the magic of a hardware RAID controller.  The operating system sees only one drive.  It should appear as I listed it above.  As an example, the biggest RAID I've dealt with consisted on 12 disks, 36GB each, in a RAID 5 configuration.  Windows 2000 A.S. listed it as "Disk 1 - Drive D : - Type = Dynamic - Layout = simple  Size = 432GB

4.  "Both machines have their independant jobs (one for web services, the other for database services)"

Then both will still need the SQL database software installed, though they won't need the exact same hardware and drive partition configuration.

5.  "If I can prove that the system is not setup correctly"

Can you try the test I've suggested where one machine gets a reboot?  You will probably want to wait until evening or whenever your traffic is lowest to do this.  In one simple stroke, this will give you all the evidence you need to go and ask them to have another look at your system.

If you don't want even a minute's downtime, the obscure error message in the hardware manager is enough to call up and ask what is up with it.

6. "Do you have any good references that I can go to for an overview on the proper configuration of a clustered server?"

No, unfortunately.  Its the kind of thing I could fiddle with if I had a manual, but not otherwise.  An additional problem is that a slick operation wouldn't even need to tell your machines that they were to failsafe for each other.  You'd just have the local router redirect packets to the other machine when one of them stopped esponding.  So, your current setup may be OK as-is.  But without SQL installed on both, I don't see how that functionality is going to be replicated.

Really, the self-induced outage by rebooting one machine is your best test and its failure your best ammunition to have a legitimate and undisputable gripe to take to the hosting company's management.

regards,
magarity
0
 
LVL 13

Expert Comment

by:magarity
ID: 6389237
Furthermore, if these machines are not yours but belong to the hosting company, I would begin to wonder if your applications do not have exclusive use.  That would help explain the hard drives not being configured as you asked.
0
 

Author Comment

by:bmccleary
ID: 6395210
Magarity,
Thank you for all of your help.  I do appreciate it.  You are correct, they are not clustered, but now at least I have a foundation to discuss this matter intelligently with the hosting company.  Take care!
0
 

Author Comment

by:bmccleary
ID: 6395213
I just tried to give you 350 points, but the server wouldn't let me give you more than 300.  So, now I don't know if you even got any points.  Let me know if you didn't.
0

Featured Post

How your wiki can always stay up-to-date

Quip doubles as a “living” wiki and a project management tool that evolves with your organization. As you finish projects in Quip, the work remains, easily accessible to all team members, new and old.
- Increase transparency
- Onboard new hires faster
- Access from mobile/offline

Join & Write a Comment

How to update Firmware and Bios in Dell Equalogic PS6000 Arrays and Hard Disks firmware update.
This paper addresses the security of Sennheiser DECT Contact Center and Office (CC&O) headsets. It describes the DECT security chain comprised of “Pairing”, “Per Call Authentication” and “Encryption”, which are all part of the standard DECT protocol.
This video teaches viewers how to encrypt an external drive that requires a password to read and edit the drive. All tasks are done in Disk Utility. Plug in the external drive you wish to encrypt: Make sure all previous data on the drive has been …
This Micro Tutorial will teach you how to reformat your flash drive. Sometimes your flash drive may have issues carrying files so this will completely restore it to manufacturing settings. Make sure to backup all files before reformatting. This w…

757 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question

Need Help in Real-Time?

Connect with top rated Experts

18 Experts available now in Live!

Get 1:1 Help Now