Replacing drives in a RAID5
- Thread starter GodAtum
- Start date
Soldato
- Joined
- 18 Oct 2002
- Posts
- 8,253
- Location
- The Land of Roundabouts
Best practise would be to backup and restore, though clearly that comes with downtime, but its the best way of safeguarding the data against corruption.
3tb isn't really a lot of data.
Admittedly iam guilty of performing the old disk shifteroo
3tb isn't really a lot of data.
Admittedly iam guilty of performing the old disk shifteroo
Associate
- Joined
- 29 Dec 2014
- Posts
- 141
- Location
- Bedfordshire
If its anything like HP kit. Once all disks are replaced you will need to go into the raid controller software and extend the array. Once this is done the extra space will be available to the OS.
Best practise would be to backup and restore, though clearly that comes with downtime, but its the best way of safeguarding the data against corruption.
3tb isn't really a lot of data.
Admittedly iam guilty of performing the old disk shifteroo
This. Especially as you're only on drive failure away from losing the data. I had a brand new WD Red fail within the first hour of use.
Permabanned
- Joined
- 9 Jun 2009
- Posts
- 11,924
- Location
- London, McLaren or Radical
Does that raid controller support expanding the array size?
Last edited:
Permabanned
- Joined
- 28 Dec 2009
- Posts
- 13,052
- Location
- london
I think raid will just reconfigure the 1.2tb as 600gb. It will do that in hp servers. You can use any disk in a raid as long as its the same RPM and larger than the other disks in the raid. If you add a disk using another slot then it will be able to expand the raid to include the disk but i think it will still only use 600gb of the 1.2tb, otherwise the other disks 600gb ones could not work as backup for the 1.2tb disk if it was to fail. Even in HP p2000 i would not recommend expanding the size of a raid by adding a disk. Not if it is production data as the performance will drop for days. It is always a better idea to create a new one if you have the slots and if you don't have the slots then build a new server and migrate the data or use backup and restore method as described earlier. If you have three slots, build a new 1.2tb raid 5 or 6 then migrate the data and then remove the old disks and then do a one time expand of the 1.2tb raid with new 3 disks etc.
Last edited:
Permabanned
- Joined
- 9 Jun 2009
- Posts
- 11,924
- Location
- London, McLaren or Radical
Sounds good... I was just checking as not all support it with ease... but looks as though that one does.
So yes, I would replace 1 drive at a time, wait for the rebuild to complete, replace the next drive, etc until they're all done & rebuilt... then you should be able to expand the array.
anything i don't mind:
You don't have to have the same RPM, for any raid array... really.
Only usually the raid performance will slow down to whatever the slowest drive can do.
So yes, I would replace 1 drive at a time, wait for the rebuild to complete, replace the next drive, etc until they're all done & rebuilt... then you should be able to expand the array.
anything i don't mind:
You don't have to have the same RPM, for any raid array... really.
Only usually the raid performance will slow down to whatever the slowest drive can do.
Permabanned
- Joined
- 9 Jun 2009
- Posts
- 11,924
- Location
- London, McLaren or Radical
Since when? I've seen multiple raid 5 arrays with drives exceeding 1TB...
A couple of years or so, longer according to some people. The amount of I/O disks have to undergo when rebuilding a RAID 5 array massively increases the chance of a URE occurring and bang goes the array. Similar when writing large amounts of data, the disks get hammered.
And add in the performance hit whilst rebuilding (drive failures do happen, especially for consumer grade disks). RAID 10 these days makes so much more sense for larger drives.
Just because you've seen them, doesn't mean they're best practice. We have a WD Sentinel at work - 4x4TB disks in RAID 5 (for some reason it won't do 10) - we've had 2 drive failures recently, WD replaced the disks quickly, but we were bricking it both times it was rebuilding - well over 24 hours each time.
And add in the performance hit whilst rebuilding (drive failures do happen, especially for consumer grade disks). RAID 10 these days makes so much more sense for larger drives.
Just because you've seen them, doesn't mean they're best practice. We have a WD Sentinel at work - 4x4TB disks in RAID 5 (for some reason it won't do 10) - we've had 2 drive failures recently, WD replaced the disks quickly, but we were bricking it both times it was rebuilding - well over 24 hours each time.
Last edited:
For about two years plus, similarly raid 50 isn't recommended for obvious reasons. Where Ive not been clear is this was mainly aimed at sata spindles. I'm personally of the opinion any raid design should consider best practices and considerations for disk failure and the impact of rebuilds and their duration. Of course R5 works and yes people use it but it is not recommended for production use. Regardless Dell should confirm for certain as per my earlier post just to be certain rather than rely for production information on here (No offence intended)
Last edited:
