On 07/22/10 16:27, roberth wrote: > Lo, > > anyone ever killed a SSD while running OpenBSD ontop of it? > > Maybe i just got a bad sample, but my Intel X25-M died after ~3 month. > Suspiscious io-error area inside /usr/src. > SMART 'End to End Error Detection count' went tits up. > Not user fixable. RMA'd just fine. > > Just wondering if someone else had any problems themself with other > SSD's running OpenBSD. > > ~roberth
I've seen failures on just about every form of non-volatile storage device I've had my hands on. It has nothing to do with OpenBSD. If you got ssd or flash or EEPROM or EPROM or ... for "reliability", you got fooled. There's a lot more to this stuff to fail than "write fatigue". Look at how a floating gate device works...it's amazing they work at all, it's amazing the expected data retention is measured in years and not minutes. Surprises me not in the least when they die from time to time... 4GB chip requires a lot more than 4 billion perfect storage elements (and line drivers and amps and ... ) The flash BIOS on my laptop died a while back. I can assure you it wasn't "write fatigue" (it hadn't been updated since it left the factory). I really doubt it was the OS. Vendor apparently knew this could happen, and provided a way for the system to re-load its BIOS from USB utilizing another chunk of (probably) flash ROM which hopefully didn't screw itself up... Low power? Flash. Fast? SSD. Reliable? gimme a two or three year old HD technology (500G SATA would probably be good 'bout now. 160G might be cutting corners to make it too cheap, 2G is too cutting edge), and a good backup system and rapid repair plan for when it fails anyway. I'm not going to swear the HD will be more reliable in statistically significant samples, but I'll spend less total time down because I'll be ready for it. Nick.
