Open LinuxETC opened 10 years ago
Interesting thought... but because cudaminer (in some situations, at least) doesn't gracefully recover from a video driver crash (often it spews repeated error messages and stops mining on all cards), there are recommendations in many of the threads I have read, advising to run an instance of CUDAminer per GPU. Here's a good example: https://github.com/cbuchner1/CudaMiner/issues/109 The above example is from windows... unfortunately...
Side question: Is there a reason this can't be done via init script?
I am hoping to have such in place via CUDAminer actually and use init.d scripts to make various calls to such more effectively. e.g., "status" (if there is an active PID file or not), "stop" (pkill PID), etc.
Just a thought and further clarification on my part is all of this request.
Would it be possible to have CUDAminer create, use, then destroy a PID lock file script (to be stored in "/var/lock" for a Linux system for example)? This would be beneficial in quite a few ways, though primarily to ensure that CUDAminer is only running one main/primary instance at a time on a system.
Thanks in advance.