Do you wonder if your IT business is truly profitable or if you should raise your prices? Learn how to calculate your overhead burden using our free interactive tool and use it to determine the right price for your IT services. Start calculating Now!
Upgrading Mirrored Hard Drives on Dell PowerEdge T110 II Server
My client has a Dell PowerEdge T110 II Server and the mirrored 320 GB hard drives are almost full. I need to purchase and install a pair 1 TB hard drives to replace the current drives.
This server has a PERC S100 (Embedded SATA Software RAID) running the 2 drives in RAID 1.
I've never upgraded hard drives in a server running RAID. If I purchase, 2 new 1 TB drives, what is the best way to clone the current mirrored drives over to the new pair of drives?
Also, I assume I can buy any brand 7200 RPM SATA HDDs as replacements, correct?
Thanks in advance.
This server has a PERC S100 (Embedded SATA Software RAID) running the 2 drives in RAID 1.
I've never upgraded hard drives in a server running RAID. If I purchase, 2 new 1 TB drives, what is the best way to clone the current mirrored drives over to the new pair of drives?
Also, I assume I can buy any brand 7200 RPM SATA HDDs as replacements, correct?
Thanks in advance.
Do more with
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
The T100 is about as low-end as RAID controllers come, so the hard drive make/model isn't terribly important, although there still could be some incompatibilities, so I'd recommend Dell-certified drives (then make/model becomes unimportant), but enterprise-class should be fine. Just don't use desktop/laptop/NAS drives like the WD Green, Blue, Red, etc.
You'll need to backup/image/clone the existing OS, replace and configure the new drives, then restore.
You'll need to backup/image/clone the existing OS, replace and configure the new drives, then restore.
The PERC won't accept a lot of drives avoid the AF (4K sectored ones because firmware won't support them) Absolutely get the Dell certified.
Best practice is to make a full bootable backup onto a scratch drive, network, or whatever. Then yank the 2 disks, put in 2 x 1TB drives, and tell the RAID controller to turn them into a mirror, and do the full sync that verifies every block. This will take hours.
Once the RAID1 has been built and zeroed, boot the recovery medium, then restore. Most of the decent backup software will resize your partitions to use the full 1TB.
If something goes bad then you haven't touched your original mirror, so you could get the controller to recognize that and try another day.
P.S. if you have the ability to hook up an external eSATA or even get a USB-3 controller for $20, then back up to that. You dont want to back up the 320GB to the network, it could take days.
Best practice is to make a full bootable backup onto a scratch drive, network, or whatever. Then yank the 2 disks, put in 2 x 1TB drives, and tell the RAID controller to turn them into a mirror, and do the full sync that verifies every block. This will take hours.
Once the RAID1 has been built and zeroed, boot the recovery medium, then restore. Most of the decent backup software will resize your partitions to use the full 1TB.
If something goes bad then you haven't touched your original mirror, so you could get the controller to recognize that and try another day.
P.S. if you have the ability to hook up an external eSATA or even get a USB-3 controller for $20, then back up to that. You dont want to back up the 320GB to the network, it could take days.
It seems like I did this same sort of thing just recently. Here are the issues that I encountered:
- IF a new hard drive was the same as the model being replaced, then it worked. Otherwise, not. The idea that one could introduce a new hard drive with *at least* the capacity of the old one didn't hold or at least didn't hold in my case. The original drives were 500GB and a new but different 500 GB model (same mfr) would not work. A (single) 1TB drive would not work either alongside the remaining 2 500GB drives.
- This one has 3 500GB in RAID 5 configuration which appears as a Virtual Disk with 930GB of space. I don't know which RAID configuration you have set up presently.
I recall that there was going to be an issue with increasing the hard drives to 1TB. There was at least a registry change as I recall to make the system compatible by enabling 48-bit LBA large-dksi support. I didn't do it but here are the instructions I found:
1) Start Registry Editor Regedt32.exe, no Regedit
2) Locate and then click the following key in the registry:
HKEY_LOCAL_MACHINE\SYSTEM\
3) On the edit menu, lick Add Value, and then add the following registry value:
Value name: EnableBigLba
Data Type: REG_DWORD
Value data: 0x1
So, if you were willing to do that much then maybe the 1TB disks would be recognized.
I do recall that it appeared to be fairly low risk / harmless but I can't prove that.
So, then maybe you'd be able to replace all the drives with 1TB drives.
And, then maybe the system would accept them if they are all the same model - the system would *not* recognize a different model 500GB drive that I tried to use.
Since I didn't do the regedit, the system saw the one 1TB drive that I swapped in as an SSD and wouldn't put it into the RAID array - and I was stuck with that. I had to stay with 500GB drives of the original model. Fortunately I found them available.
I came away regretting that I'd tried to upgrade old hardware because the issues I ran into would otherwise be long past us and I was learning things I'd rather not have to deal with - because it was out of date.
- IF a new hard drive was the same as the model being replaced, then it worked. Otherwise, not. The idea that one could introduce a new hard drive with *at least* the capacity of the old one didn't hold or at least didn't hold in my case. The original drives were 500GB and a new but different 500 GB model (same mfr) would not work. A (single) 1TB drive would not work either alongside the remaining 2 500GB drives.
- This one has 3 500GB in RAID 5 configuration which appears as a Virtual Disk with 930GB of space. I don't know which RAID configuration you have set up presently.
I recall that there was going to be an issue with increasing the hard drives to 1TB. There was at least a registry change as I recall to make the system compatible by enabling 48-bit LBA large-dksi support. I didn't do it but here are the instructions I found:
1) Start Registry Editor Regedt32.exe, no Regedit
2) Locate and then click the following key in the registry:
HKEY_LOCAL_MACHINE\SYSTEM\
3) On the edit menu, lick Add Value, and then add the following registry value:
Value name: EnableBigLba
Data Type: REG_DWORD
Value data: 0x1
So, if you were willing to do that much then maybe the 1TB disks would be recognized.
I do recall that it appeared to be fairly low risk / harmless but I can't prove that.
So, then maybe you'd be able to replace all the drives with 1TB drives.
And, then maybe the system would accept them if they are all the same model - the system would *not* recognize a different model 500GB drive that I tried to use.
Since I didn't do the regedit, the system saw the one 1TB drive that I swapped in as an SSD and wouldn't put it into the RAID array - and I was stuck with that. I had to stay with 500GB drives of the original model. Fortunately I found them available.
I came away regretting that I'd tried to upgrade old hardware because the issues I ran into would otherwise be long past us and I was learning things I'd rather not have to deal with - because it was out of date.
Respectfully, a lot of problems with the previous post. (First, i have developers NDA, so am quite familiar with the firmware).
1. The controller couldn't care less about make / model. There is NOTHING in the firmware that looks at that. If a disk isn't supported, it is going to be for other reasons.
2. The 48-bit LBA registry setting makes zero sense. The controller makes the logical volume emulate a SCSI drive. 48-bit LBA is applicable to the ATA/SATA instruction set, not the SCSI instruction set.
3. As i wrote before, the biggest incompatibility issue is if you try to use a disk with anything but 512 byte sector size; or if total # of sectors > 2^32 -1, or if doing something other than RAID1 and the total number of usable logical blocks is > 2^32 -1 .
1. The controller couldn't care less about make / model. There is NOTHING in the firmware that looks at that. If a disk isn't supported, it is going to be for other reasons.
2. The 48-bit LBA registry setting makes zero sense. The controller makes the logical volume emulate a SCSI drive. 48-bit LBA is applicable to the ATA/SATA instruction set, not the SCSI instruction set.
3. As i wrote before, the biggest incompatibility issue is if you try to use a disk with anything but 512 byte sector size; or if total # of sectors > 2^32 -1, or if doing something other than RAID1 and the total number of usable logical blocks is > 2^32 -1 .
Thank you all for the feedback.
I'm almost certain this is wishful thinking, but since this is a RAID 1, would be possible to replace just one of the 320 GB drives with a blank 1 TB drive and have the RAID automatically rebuild the mirror copy on to the blank 1 TB drive. Then once that would be done, drop in the 2nd blank 1 TB drive and repeat the rebuild process?
I'm almost certain this is wishful thinking, but since this is a RAID 1, would be possible to replace just one of the 320 GB drives with a blank 1 TB drive and have the RAID automatically rebuild the mirror copy on to the blank 1 TB drive. Then once that would be done, drop in the 2nd blank 1 TB drive and repeat the rebuild process?
Since this is a RAID 1 configuration, would be possible to replace just one of the 320 GB drives with a blank 1 TB drive and have the RAID automatically rebuild the mirror copy on to the blank 1 TB drive. Then once that would be done, drop in the 2nd blank 1 TB drive and repeat the rebuild process?
Once you do, your VD will be the same size, and the S-series controllers do not have any way of expanding the VD to fill the new drives. That's why backup/restore is your only option here.
Thanks. What about if I used a program like Gparted running from a Linux boot disk to expand the partition size from 300 GB to 1 TB?
Would the controller allow/accept/recognize that??
Would the controller allow/accept/recognize that??
No, this isn't a question of partitions. We are talking about the Virtual Disk size, which is essentially the "disk" size that is being made available to the OS. The equivalent on a desktop would be upgrading a 500GB disk to a 1TB disk using Gparted instead of replacing the disk with a larger one. The VD is the "disk" that the OS can see, so you must make the VD larger before the OS can see it (THEN that is when partitioning happens). Third-party utilities can't expand a RAID VD - only extend the partitions afterwards.
dlethe: I don't understand the tone of your response.
It may well be that with your better knowledge of the firmware that you have good information.
However, I find your first assertion a bit shocking:
I purchased a new 500GB Western Digital hard drive model WD5003AZLX and used it to replace a failing WD5003ABYX 500GB HD. The system recognized it as an SSD and it was unusable in the RAID system.
I purchased a new 500GB Western Digital hard drive model WD5003ABYX with the model number matching what was being replaced. This worked.
I realize that this is empirical data and only a single data point. Yet, that's what happened.
Perhaps it would be good for others to know why? I sure don't.
As far as the 48-bit LBA: as I said, I didn't try it. These ARE SATA drives and I can only conclude that it's necessary for them to be recognized in the system properly before the SCSI emulation can happen. The RAID system will not accept what the system sees as an SSD. They aren't recognized in the system properly without the registry edit as I understand it. It's a topic that's easy to research if necessary.
It may well be that with your better knowledge of the firmware that you have good information.
However, I find your first assertion a bit shocking:
1. The controller couldn't care less about make / model. There is NOTHING in the firmware that looks at that. If a disk isn't supported, it is going to be for other reasons.Either the firmware you refer to is different that the one I encountered or what I did was what you would call "unusual". I don't think it was. Here's what I did:
I purchased a new 500GB Western Digital hard drive model WD5003AZLX and used it to replace a failing WD5003ABYX 500GB HD. The system recognized it as an SSD and it was unusable in the RAID system.
I purchased a new 500GB Western Digital hard drive model WD5003ABYX with the model number matching what was being replaced. This worked.
I realize that this is empirical data and only a single data point. Yet, that's what happened.
Perhaps it would be good for others to know why? I sure don't.
As far as the 48-bit LBA: as I said, I didn't try it. These ARE SATA drives and I can only conclude that it's necessary for them to be recognized in the system properly before the SCSI emulation can happen. The RAID system will not accept what the system sees as an SSD. They aren't recognized in the system properly without the registry edit as I understand it. It's a topic that's easy to research if necessary.
I thought the "Virtual Disk" merely represents the 2 drives (currently 300 GB each) mirrored in RAID 1 configuration.
So there is no Dell-provided RAID software utility that allows dynamic copying and expanding of the size of the virtual disk?? Assuming, of course, that there is now a much larger physical disk attached to the RAID controller card).
So there is no Dell-provided RAID software utility that allows dynamic copying and expanding of the size of the virtual disk?? Assuming, of course, that there is now a much larger physical disk attached to the RAID controller card).
You replaced a WD RE4 (RAID Edition) with a WD Black. Black is designed as a desktop/laptop consumer-grade drive and should not be used in RAID arrays or servers. Non-certified, non-enterprise grade drives can display a number of quirks because of incompatibility.
The Virtual Disk manages two separate disks in a mirrored configuration on the hardware/controller side, but it presents the single Virtual Disk it manages to the OS to use. The OS is oblivious to number of disks in the array or that there is even an array present.
Only Dell software can speak to the controller to query what devices it is managing. This is ABSOLUTELY a supported feature of Dell controllers, but the S-series controllers are the bottom-of-the-barrel when it comes to controllers and do not support expansion in this way. The H-series controllers do. I would always recommend the H7x0 controller for any application. Even Windows mirroring would be preferable to the S-series controllers.
The Virtual Disk manages two separate disks in a mirrored configuration on the hardware/controller side, but it presents the single Virtual Disk it manages to the OS to use. The OS is oblivious to number of disks in the array or that there is even an array present.
Only Dell software can speak to the controller to query what devices it is managing. This is ABSOLUTELY a supported feature of Dell controllers, but the S-series controllers are the bottom-of-the-barrel when it comes to controllers and do not support expansion in this way. The H-series controllers do. I would always recommend the H7x0 controller for any application. Even Windows mirroring would be preferable to the S-series controllers.
PowerEdgeTech: do you mean me?
You replaced a WD RE4 (RAID Edition) with a WD Black.I replaced a WD5003ABYX with the same WD5003ABYX. I'm not aware that there are different models with the same model number. So I'm confused. It would be nice to know.
I thought it was the author, but I was addressing your comment:
I purchased a new 500GB Western Digital hard drive model WD5003AZLX and used it to replace a failing WD5003ABYX 500GB HD. The system recognized it as an SSD and it was unusable in the RAID system.You said you used a WD5003AZLX (WD Black https://www.newegg.com/Product/Product.aspx?Item=N82E16822236345) to replace a failing WD5003ABYX (WD RE4 https://www.amazon.com/Enterprise-Drive-Cache-WD5003ABYX-Model/dp/B003SALVN4), which you said didn't work. You said it worked to replace the RE4 with another RE4, which makes sense. The RE4 is not certified by Dell, but the older RE4's were validated by WD on Dell controllers. You said you didn't know why it didn't work ... now you do :)
I purchased a new 500GB Western Digital hard drive model WD5003ABYX with the model number matching what was being replaced. This worked.
I realize that this is empirical data and only a single data point. Yet, that's what happened.
Perhaps it would be good for others to know why? I sure don't.
Fred Marshall - My response was simply to correct the mistakes in your post. I sincerely apologize as clearly interpreted my response as something that had a "tone". I'm just blunt, and don't read anything else into it other than that.
It looks like PowerEdgeTech handled the reasons nicely on why Black vs RE4 change.
As for the 48-bit thing, the controller presents the LUN as a SCSI device. The O/S does not see individual disks, and certainly not ATA/IDE/SATA drives.
P.S. USB sticks are the same way, they emulate SCSI devices.
It looks like PowerEdgeTech handled the reasons nicely on why Black vs RE4 change.
As for the 48-bit thing, the controller presents the LUN as a SCSI device. The O/S does not see individual disks, and certainly not ATA/IDE/SATA drives.
P.S. USB sticks are the same way, they emulate SCSI devices.
Premium Content
You need an Expert Office subscription to comment.Start Free Trial