Hi there,
I have found strange behavior for my mixed rigs. It works quite good for at least 1-2 days, but got a lot of invalid shares (nvidia gpus mainly) after then. It took 2 days, 3 days or more often even 4 days without any invalid share and after that fall down to approximately 60% efficiency (valid/invalid shares for the whole mixed rig). After reboot it works as before, no invalid shares for several days and then the scenarios repeats again:
Any hint please?
Have similar problem. For me I think its mostly due to rebuilding DAG with the OC settings turned on. I would love to see a WatchDog that restarted the miner after certain invalid shares.
I have the same issue with both AMD and Mixed rigs. They all work perfect for about 3 days, but then start to generate the occasional invalid share on the worker page. Those invalid shares are not listed on my miner statistics page. A reboot fixes the issue for another 3 days.
Shouldn’t we ask for a new hiveos watchdog feature then?
I have now switched to t-rex miner and executing a script (reboot script) if the dag is rebuilding. The t-rex miner also has much better feature for building dag for RTX-cards so for me the problem is basically solved now.
I have set dual miner (phoenix - AMD, T-rex - Nvidia) setup for the purpose and changed T-rex config extra arguments as follows:
“dag-build-mode”:“2”
“script-epoch-change”:"/sbin/reboot"
Everything works now regarding the invalid shares.
Unfortunately the miner hashrate reporting is not working as expected now at the pool:
The reported hashrate is invalid and changes during the period somehow random. I checked phoenix reporting is not working as expected. But i didn’t touch phoenix config at all. Any hint regrading reporting hashrate using dual miner config?
The calculated hasrate is ok.
Thanks.
Im about 99% certain its new epoch.
Gotta create new DAGs, which breaks them if done while clocked.
As we already have the delay on OC settings, we really just need a setting in watchdog that restarts the miner when the epoch steps up.
Frustrating, as it’ll chug along for hours to days with 99.99%+ accepted, then just faults hundreds in a hurry. Threw almost 1000 down the well before i caught it.
Came here looking for at least a workaround.
Normal watchdog doesnt catch it, as its looking for reported hashrate, which doesnt really change when this happens
Did you solve this? I have the same problem. Teamredminer and Amd cards. Every time when DAG changed (every 4 days)
I have solved it somehow for t-rex:
I do not know if teamred has such parameter to trigger something when epoch changed.
If I updated t-rex above 0.21.4, I have an issue starting the miner at all. It says something about wrong command line parameters…
This topic was automatically closed 416 days after the last reply. New replies are no longer allowed.
