3

I have a Dell PowerEdge R300 with a Dell SAS 6/iR Internal Controller RAID PCIe and two 250GB SATA HDDs in RAID1 configuration. One of the HDDs has died so I want to swap it out for another. At the same time I'd like to increase storage to 500 or 750GB. Can I just plug in say a 500GB HDD, wait for it to sync, then swap out the other old drive for another 500GB drive?

Or, how do I find out if this is possible - i.e. what's the terminology for this if I want to search around in manuals etc?

Also, I'm assuming that I can get any 3.5" SATA (or SATA II) drive and it'll work and be hot-pluggable. Is this correct?

many thanks!

Rory
  • 482
  • 5
  • 12
  • 22

2 Answers2

3

The feature term you are looking for is "Online Capacity Extension". Dell SAS iR controllers do not support it, the PERC and CERC controllers do. So with the iR controller, you would have to insert your two new disks, create a new large container and copy the data over from the old one.

Also, I'm assuming that I can get any 3.5" SATA (or SATA II) drive and it'll work and be hot-pluggable. Is this correct?

Basically yes, although controller-specific limitations may apply (like a 2TB volume limit). You should make sure that you use "24/7" rated drives - the appropriate models also usually come with larger caches, higher MTBF ratings and a longer manufacturer warranty (3 to 5 years).

the-wabbit
  • 40,319
  • 13
  • 105
  • 169
2

  1. When you replace disks one by one and sync, you'll end up with a new RADI1 array, with the size of 250Gb, and the rest of the space on the disks will be unused, but available, for creating another raid array on the rest of the space available. If you want the entire space in a single radi1, you need to backup, test backup, create a new large array, and restore the system to the new array.

  2. Only Dell HDDs will be supported in a Dell server. They are the onyl drives that were tested to work in this machine, and if something goes wrong, support will ask you tyo remove any non-Dell kit before proceeding with testing your machine for errors. Obviously, without drives, it will not be a simple task to reproduce most of the issues that can occur.

dyasny
  • 18,482
  • 6
  • 48
  • 63
  • You can do this dynamically with PERC raid controller cards, but definitely not with the SAS 6/ir controller card. – Bryan Jan 26 '12 at 12:35
  • Since when? the 5i and 6i definitely couldn't do this. Maybe it was added in H700, but I'm not aware of such a change – dyasny Jan 26 '12 at 13:05
  • I've not done this myself, but a colleague has. A couple of [Other](http://serverfault.com/questions/270897/upgrading-raid-5-array-drives-in-a-dell-2950-with-perc-6i-controller) [questions](http://serverfault.com/questions/208825/how-do-you-expand-a-raid-disk-array-in-a-dell-2850) here on SF cover this. – Bryan Jan 26 '12 at 14:12
  • could be, could be. I haven't touched the 11G cards much since I left Dell – dyasny Jan 26 '12 at 15:54
  • 3
    `They are the onyl drives that were tested to work in this machine` - nah, they are the only drives where you have paid a 75% extra for support. Basically, Dell uses commonly available disks (usually Seagate) and occasionally interchanges manufacturers and models when delivering replacement parts. I wonder if any support engineer is going to cross-check the drives' serial numbers against the Dell database upon opening of a support case. – the-wabbit Jan 26 '12 at 17:12
  • Actually, yes and no. I used to be one of those engineers, and finding out whether or not the drive is authentic is very easy, takes just a few seconds, because when a case is opened, one looks at the server information anyway, and the disk info is right there in front of you. Besides, I've seen quite a few cases of really weird disk behaviour when non-dell disks were used. Once the customer put the original drives back in everything just worked. So... I agree, the price diff is unreasonable, but it's cheaper than downtime and troubleshooting hours – dyasny Jan 26 '12 at 17:39
  • I would agree that a case with *Samsung Spinpoint" drives in a Poweredge might look more than suspicious. But how do you tell if a specific Seagate Cheetah or Constellation has been Dell-branded or not? The firmware or identification strings are left unchanged. – the-wabbit Jan 26 '12 at 18:31
  • disk size, make and sometimes firmware versions are documented. if you buy a server with no disks and then call in about a disk issue, it's kinda obvious, isn't it? Same goes for appearing to have small drives, and calling in and providing logs and reports from larger drives. Besides, firmware versions are well known, if you have anything that is not in the list - busted. There are loads of ways, and it's perfectly fine to use your own hardware, as long as you don't expect it to be supported. – dyasny Jan 26 '12 at 18:52
  • @dyasny I do not know if Dell is an exception here, but as soon as parts can be purchased through channel distribution (and I believe Dell is allowing for this for a couple of years now), a support technical would not be able to rely on the original configuration documentation. The user might have bought branded drives and inserted them into the machine, you just would not be able to tell without querying a database of all dell-branded drives for serial numbers. – the-wabbit Jan 27 '12 at 11:06
  • When I was still working there, and that was over 3 years ago, server changes were meticulously updated. The only discrepancies I've seen were in countries where there is no Dell presence at all, and even then it was very easy to see if a drive was branded or not – dyasny Jan 27 '12 at 11:15