Er, whoops?
This evening, one of the hard disks on Pell started spitting out nice elaborate scsi errors about a block conveniently located in the middle of one of the disks. Er, oops? This is very inconvenient, because, unlike all of my other servers, pell is a pure scsi setup (fast wide scsi, driven by a BusLogic flashpoint card that I got back in the old days when lnz (RIP) was writing the BusLogic (also RIP) flashpoint (also also RIP) device driver. Anyway, these disks are about 7 years old now and I'm not exactly sure where I could go to get new 4gb scsi disks that are reliable enough to run for 60,000 hours before they start to climb into the handbasket.
I suppose I could always spend $35 and buy a 160gb IDE drive, but I'm not sure how the 2.0.28+orc kernel that Pell is running would react when confronted with a disk that has a seriously NON-32bit geometry. Sigh. What a pain.