Quote from: Alan Lupton on July 17, 2011, 09:39:02 PM

Quote from: jondecker76 on July 17, 2011, 04:06:55 PM

I want to gather suggestions on the preferred cli email program to be used by smartcoin. I'll go with the most requested.

I finished a new routine which monitors for GPU lockups (the kind common when pushing overclocking a bit too far)
All of my tests so far indicate that a full reboot is needed to recover from locked card conditions. Currently, the routine will monitor for this kind of hard lock condition, and if found it will Log and error, shut down smartcoin and fire off an email. I'm going to do some tests to see if its possible to revive a locked card, but until then this is about as elegant as I can come up with.

Any further suggestions on this?

Why would you kill smartcoin and kill the other working devices? Let's say I go to bed and sleep 7 hours. After 1 hour one GPU locks. Then you kill smartcoin and 6 hours idle time for 3 other GPUs. That doesn't make sense to me. The current state is even better, because when getting up I can easily detect the the lockup by looking at the stats and I wouldn't waste 6 hours x3 GPUs hashing time.

Email, wonderful. Shutdown smartcoin, very bad idea imo.

I have a GPU that locks up if I OC over about 960 MHz (it's a 5830). When it locks up, the other GPU often stops processing as well.

Actually, Jon, what I'd prefer is that you can specify a script to run when a card locks up. I can then do whatever I want. Something like "lockup.sh". In my script I would:

1) Capture Temps
2) Capture GPU Status (overclock rates, etc)
3) Send the above to me in an email
4) Reboot the box

By having a script, I could even code whether or not I want to reboot. Maybe setting a flag each time it reboots and if I see it reboot 4 times unattended, I just leave it down (because I need to look into what's going on).

Thoughts?