On 04/25/2013 02:58 PM, Konstantin Olchanski wrote:
>>     joe@george:~$ sudo smartctl -iA /dev/sdd
>>     Model Family:     SandForce Driven SSDs
>>     Device Model:     Corsair CSSD-F240GB2
>>     Firmware Version: 2.0
>>     User Capacity:    240,057,409,536 bytes [240 GB]
>>
>>     171 Program_Fail_Count      0x0032   000   000   000 Old_age Always       -       1
>>     174 Unexpect_Power_Loss_Ct  0x0030   000   000   000 Old_age Offline      -       45
>>     181 Program_Fail_Count      0x0032   000   000   000 Old_age Always       -       1
> Your SSD is toast... Program_Fail_Count non zero is bad. Unexpected power loss count non zero is bad.
>
I agree, it's now two days with Einstein at Home running using 100% CPU 
and GPU with no crashes.

The symptoms were certainly strange.  To recap:

On average once or twice a day the system would just stop working with 
error messages occasionally and complete freeze most of the time.  A 
power cycle seemed to reset the clock and let me work for another 4 to 8 
hours.

Error messages included "too many files open", "i/o error",  "bus error" 
and probably more.

The SSD was my system disk but swap, /tmp, /usr were on spinning media.  
When it worked it did everything I tried, copy files and diff,  read all 
files, run SMART diagnostic tests.

The SSD was only marginally faster (judging by response time).  I don't 
think I'm going to replace it with another SSD.  However, I have nothing 
against SSD, almost 2 years of trouble free operation and a sample size 
of one does not lead me to any conclusions about them.

Thanks to the list, I now have more Linux diagnostic tools.

Joe