I use HiveOS for mining ALPH.
The miner sometimes gives error because of overclock. The miner restarts mining progress. I dont want its restart. I want to see the error which gpu has.
How could it possible ? is there any code to write in extra config ?
I see the error code if i only watch the minnig progress...
But there are so many rigs. ı cannot watch the progress. I only need the information which gpu has the problem.
20220916 14:40:20 [ OK ] 40/40 - 14.44 GH/s, 55ms ... GPU #0
20220916 14:40:33 [ OK ] 41/41 - 14.44 GH/s, 47ms ... GPU #2
20220916 14:40:41 TREX: Can't find nonce with device [ID=4, GPU #4], cuda exception: CUDA_ERROR_LAUNCH_TIMEOUT, try to reduce overclock to stabilize GPU state
20220916 14:40:41 WARN: Miner is going to shutdown...
20220916 14:40:42 Main loop finished. Cleaning up resources...
20220916 14:40:42 ApiServer: stopped listening on 127.0.0.1:4059
20220916 14:40:44 T-Rex finished.
t-rex exited (exitcode=0), waiting to cooldown a bit
Trying to release TIME_WAIT sockets:
tcp 0 0 127.0.0.1:33498 127.0.0.1:4059 TIME_WAIT
tcp 0 0 127.0.0.1:33478 127.0.0.1:4059 TIME_WAIT
tcp 0 0 127.0.0.1:33494 127.0.0.1:4059 TIME_WAIT
tcp 0 0 127.0.0.1:33482 127.0.0.1:4059 TIME_WAIT
tcp 0 0 127.0.0.1:33490 127.0.0.1:4059 TIME_WAIT
tcp 0 0 127.0.0.1:33486 127.0.0.1:4059 TIME_WAIT
20220916 14:40:48 T-Rex NVIDIA GPU miner v0.26.5 - [Linux]
20220916 14:40:48 r.b70faa49dd35
20220916 14:40:48
20220916 14:40:48
20220916 14:40:48 NVIDIA Driver v510.60.02
20220916 14:40:48
20220916 14:40:48 + GPU #0: [00:03.0|2204] GeForce RTX 3090, 24268 MB
20220916 14:40:48 + GPU #1: [00:04.0|2204] GeForce RTX 3090, 24268 MB
20220916 14:40:48 + GPU #2: [00:05.0|2204] GeForce RTX 3090, 24268 MB
20220916 14:40:48 + GPU #3: [00:06.0|2204] GeForce RTX 3090, 24268 MB
20220916 14:40:48 + GPU #4: [00:08.0|2204] GeForce RTX 3090, 24268 MB
20220916 14:40:48 + GPU #5: [00:09.0|2204] GeForce RTX 3090, 24268 MB
I take it your running HiveOS? Let TRex do a log file and in the directory where the log file is do
grep -i 'CUDA_ERROR_LAUNCH_TIMEOUT\|nonce' and it should show you which gpu is at fault.
Windows stops the mining progress if any error happens. It is so simple. I just add "pause" word to the next line in the BAT file.
I want to know how to do it the same thing in HiveOS. I dont want to save all logs because I use flash disks on 61 rigs.
By the way, in ethash algorithm: The rig reboot at any error. But HiveOS shows a notification about it.
In blake3-alph and in kawpow: The miner restarts without any notification about it.
So, I ask if it is possible the same thing with other algorithms as ethash do.
I don't know if hiveos is also writing errors of gpu cards to syslog which is normally used on linux systems.
You can find it /var/log/syslog. This logfile adheres to the logrotate file implementation of linux.
Try with the sugestion I mentioned before for the logfile.
Looking at t-rex wiki on github t-rex itself doesn't have many log options like some other miners do like curtail the size of a logfile.
I suggest make an issue at t-rex github and ask for a logrotate option or a size limiting logfile option.
Otherwise maybe the webui of t-rex might be a possibility.
Regretfully PhoenixMiner looks to be abandoned but it has an option for logsmaxsize to avoid large logfiles.
