On 7/16/2017 4:24 PM, Ben Hutchings wrote:
Have you seen multiple crashes with a warning logged referring to the same file (fs/dcache.c)? Or do you see different filenames or messages every time?
I honestly don't recall and grepping through my syslogs indicates that I don't have any records. This is because the system wasn't able to write to disk on previous events.
I configured the sysrq key sequence so that I can reboot it over an IP KVM when this happens, but I'm not able to get any useful info out of it most of the time. Very frustrating.
"Optiplex 3020" appears to refer to two desktop models that are both physically small and possibly not designed for this number of drives. Are you sure that there is sufficient air flow to cool the hardware (CPU, motherboard, drives)?
This one is the larger tower chassis. I do not suspect any thermal problems because I actively monitor this system resources and have looked at historical data and didn't notice anything that I could blame on environmental factors. I use Cacti (one-minute datapoints) and then have a script that dumps all kinds of common system commands, including sensors, and they don't show anything unusual.
The only pattern I have noticed was that the system would lock up during a backup to it's USB3-attached disk drive. Same time each night. My graphs show no thermal changes caused by the backups.
Just keep an eye out for anything similar. Thanks. I'll try 4.12 when it's out here soon.