Woke up around 0315 with horrible sinus crud but no sense of impending doom.
Turned on monitor; wiggled mouse.
The NoMachine console for the server had come disconnected...?
OK, so switch the keyboard & monitor over to the server's physical console (Scroll Lock, Scroll Lock, 1, and push the second button from the bottom on the left side of the monitor until it selects the server's VGA signal).
Read error on swap device? Uh-oh.
Hit RESET and wait for it to boot.
All filesystems check OK.
With everything back up, look at the GUI disk utility thing. It says all the SMART statistics are OK. Drive has been powered on for 2.7 years. No uncorrectable read errors reported. So was the kernel oopsie a hallucination of some sort?
Run short-form self-test: all OK.
Now running the long-form self-test.
Meanwhile, there's a spot of confusion that needs sorting out for a client. At least all the computers are working for the moment.
Update: The long-form test completed uneventfully. Guess the drive'll hold on for a while yet. I'll probably be migrating a goodly chunk of this server's functions over to Linode during the end-of-year dead time.
Update 2: Whoops! The second drive, meanwhile, with 3.7 years on the clock, is accumulating a rather alarming reallocated-sector count. That's the bulk-data drive. So maybe I'd better order a replacement for it, and plan on transplanting the data... and, meanwhile, make sure I have good backups.
Update 3: Grrr. I look at 2TB drives... and most of the ones that turn up with good reviews are optimized for RAID use. A notable exception is the WD "Green" flavor, which was actually the first thing that came to mind, but apparently it has an unfortunate habit of parking its heads after 5 seconds of inactivity, which, especially in a Linux environment, can lead to rapid exhaustion of the finite number of park cycles before the drive fails. So, not ordering a replacement drive right now; More Research Needed. But soon. (Also, I need to refresh my memory as to what speed of SATA the server's motherboard supports; I know for certain that some SATA I host interfaces and some SATA II drives just don't get along at all, resulting in system bus wedgies, so there's some question as to how well this machine would work with a shiny new SATA III drive.)
Update 4: Oh, great. Looking at Newegg and Amazon, for drives with good average ratings... I'm seeing about 20% of the reviews being 1 star, with a general theme of "DOA! Do not buy!" Then some of the other reviews mention the drive failing after a few months or a year. Well. It's looking like a brand-new drive, even an "enterprise" model with a 5-year warranty, may have a shorter life expectancy than my current drive with 3.7 operating years on the Calvinometer and warning signs of impending failure. Maybe I need to set up a RAID after all. (More drives, more power consumption: foo.)
Update 5: Vendor responses to the one-star reviews consistently appear to express the sentiment of God's final message to His creation, but I rather suspect the true meaning is closer to the upper half of "Share and Enjoy!"
Update 6: I ended up ordering a 2TB WD "Green" drive from Amazon; should be here Thursday. Meanwhile, I also downloaded the WDIDLE3 program, found this handy tool for creating a bootable USB thumb drive (even comes with its own copy of FreeDOS, so there's no need to download a bootleg set of Windows 98 boot-disk files), used it to format a 32MB thumb drive, copied WDIDLE3 onto it, and confirmed that it boots... so, when the drive arrives and I get around to installing it, I'll be able to set the head-parking time to something reasonable, like 3 minutes instead of 5 seconds.
Comments