Replacing server hard drive best practice...?

[Koldov]

I've returned back to school after the summer to find an error reported on the server.

 

"Predictive Failure reported.Part Number = CN0W347K7262247R03XAA03: Physical Disk 0:1:5 Controller 0, Connector 0"

 

OMSA lists it as non-critical, but it obviously needs replacing.

 

It's an old DELL T420 (tower) and I think I've worked out the disk is one of the 600GB (SAS 6, 15K, 512b, 3.5 inch Hard Drive on a PERC H310 Adapter).

 

The current set-up is 2x300GB RAID 1 (OS) and 4x600GB RAID 10 (DATA)

 

The server is out of warranty, so I've contacted DELL who have quoted me £86.90 for a replacement (inc VAT and delivery)... this sounds very cheap to me, so I presume it is a refurb... and it only has a 90 day warranty from what I can see.

 

I can't remember what I've done before, so I just wondered if I should do anything other than take it offline in OMSA and replace it 'live' and let it rebuild the RAID overnight (not sure if it is hot-swappable or where to check)?

 

Or can/should this be done in the RAID config or iDRAC on a reboot?

[Norphy]

Are the current hard drives in hot-swappable caddies or do you have to take the lid off the server to get at the hard drives?

 

If the drives are in caddies, just swap the drives while the server is on and let the controller handle the rest. That will be fine. If you need to take the lid off the server to get at the drives, you'll have to schedule some downtime and turn the server off.

[Koldov]

Not 'lid off' as such, but removing the front cover gives me this view...

 

 

I think they are in caddies?

[supportman]

Far better idea these days to virtualize your servers and back them up as VM's. Just restore to a server with a good storage volume.

 

Risky hard drive replacements on live servers should really not be a thing these days.

[Koldov]

Thanks that is a good idea!

 

If I had more than one server I would definitely do that...

[Davit2005]

Not 'lid off' as such, but removing the front cover gives me this view...

 

[ATTACH=CONFIG]62940[/ATTACH]

 

I think they are in caddies?

 

Yes those are Caddies, push the red button on the caddies and will release the handle on the caddie to be able to pull the drives out. I'm pretty sure those are Hot-Swap normally on those Dells you only need to power down to add a virtual disk i.e. new raid disk group etc.

 

https://i.dell.com/sites/csdocuments/Shared-Content_data-Sheets_Documents/en/uk/Dell_PowerEdge_T420_Spec_Sheet.pdf

[jmak]

Thanks that is a good idea!

 

If I had more than one server I would definitely do that...

I used to have a single physical server (Dell T410) that I virtualized and split a few roles out. As well as proper backups (professional solution) I used to do weekly backups of the whole VM to disks on-site. No extra actual resilience as I couldn't run any kind of cluster or hot-swap setup, but I reckoned I'd massively increase my chance of getting up and running without any outside assistance and in any case be much quicker in than our external support rebuilding from backups.

 

I had two DCs on the same physical machine which meant I could update and install stuff on one of them without affecting service. Once I had most of the data in O365, I reckon I could actually have got most things running from hyper-v in Windows 10 on a desktop PC. I originally virtualized using sysinterals disk2vhd.

 

Anyway - good luck with your disk replacement.