| 1 |
"Mark Haney" <mhaney@××××××××××××.org> posted |
| 2 |
493597DE.7080201@××××××××××××.org, excerpted below, on Tue, 02 Dec 2008 |
| 3 |
15:17:34 -0500: |
| 4 |
|
| 5 |
> I certainly do appreciate the offer, but afer more digging, I'm pretty |
| 6 |
> well convinced the HDD is going bad quickly. Fortunately, it's not a |
| 7 |
> critical server (for now) and I'm glad this happened BEFORE we went live |
| 8 |
> with the DNS slave server. |
| 9 |
|
| 10 |
Ouch! |
| 11 |
|
| 12 |
For production use, I'd definitely consider at least a 2-disk RAID-1 or |
| 13 |
perhaps a 4-disk RAID-6 if you want two-way check-sums, for that very |
| 14 |
reason. With disks what they are now ($30 US shipped, SATA 40GB, |
| 15 |
pricewatch.com, less if you'll take pulls tho I wouldn't), even if a new |
| 16 |
server can be imaged and brought on line quickly, it's just not worth the |
| 17 |
hassle, for the cost of switching out a single disk and continuing in |
| 18 |
place. Get the mdadm disk failure and walk over and hot-swap, then |
| 19 |
monitor it while doing other things until the RAID rebuild is complete. |
| 20 |
Pretty hard to beat. |
| 21 |
|
| 22 |
-- |
| 23 |
Duncan - List replies preferred. No HTML msgs. |
| 24 |
"Every nonfree program has a lord, a master -- |
| 25 |
and if you use the program, he is your master." Richard Stallman |
Archives