Many of my customers still run older Evolution servers, but are not refreshing the hardware due to the impending migration to the Asure Evolution cloud.
The most common hardware failures, by far, are the hard drives, and with lapsed warranties, I typically send my customers to third-party sellers of spare drives to avoid paying through the nose with Dell. I have been very happy with ServerSupply.com, but customers have used eBay as well.
These drives only have to last another year or so; I've had very good experience with refurbished drives, no need to purchase a brand new one.
Important: if ordering a spare drive for a Dell server, it's important to look for "Dell OEM" in the product description; a stock drive from Seagate or Western Digital or whatever may not work in some Dell RAID controllers unless it has Dell-specific firmware. You can try a stock drive, but I don't recommend it unless you're sure you don't have one of the brand-enforcing RAID controllers.
In any case, I usually recommend that customers get more than one drive so they're covered in advance for the inevitable next failure (especially for the Linux database servers that use 6 or 8 identical drives), and that they follow this procedure for doing the physical drive replacement. This mostly applies to warranty replacements too.
-
You'll need a screwdriver, a Sharpie marker, and (if you bought a second spare drive) some blank Post-It notes.
-
Be really sure you have identified the right server; I've had a customer pull a good drive from a good server because the machines weren't marked properly.
-
Be really sure you have identified the failed drive; it's usually lit with a red or amber light, and if I have told you a drive has failed, I'll have told you the slot that it's in. Note: drive slots are usually usually numbered from 0, not 1.
-
Pull the failed drive from the machine — this can be done while it's running, that's why we have RAID — and immediately put a Post-It note on it with a big black X in Sharpie marker: it's surprisingly easy to get confused about which drive is which. Put the bad drive aside.
-
On the replacement drive, write in Sharpie "Installed <DATE>" on the label in case there is ever a question about the history of this drive.
-
Remove the old drive from the carrier (that's what the screwdriver is for), insert the replacement drive into the carrier. Pay attention to the orientation so it goes in the same way as the one that came out. It's not easy to get them backwards, but I'm sure it's possible.
-
Insert the new drive back into the server, insuring it's firmly seated. It should start rebuilding the array immediately with a lot of drive activity.
Note: if the drive lights don't come on within a few seconds, it's not in right.
-
If you bought a spare drive, write "Spare for <SERVER>" on the Post-It, stick it on the spare, and put it near the machine. You will thank yourself in six months when another drive fails and you'll know right away whether the drive in your hand works for that machine.
-
If I'm your consultant, let me know you did this so I can keep an eye on the array rebuild remotely.
-
Shred or otherwise dispose of the failed drive
Note: if this drive was replaced under warranty, almost all vendors require that you return the bad drive with the provided label and box. Be sure to do this to avoid being charged for it.
Comments