Open xmanzero opened 6 years ago
In my experience every miner that does that eventually crashes the drivers, prefer when I'm not looking anyway that nothing be polling the GPU besides the mining.
Then again I got crashes using AB while miners are running too, so I basically don't care about any of these stats they are too 'expensive' when they break mining (curiosity killed the minercat)
I think this has been discussed quite a few times in the past by fireice and the reason it hasn't been added is because he wants to keep the dependencies to a minimum. Adding more dependencies just increases the chances of there being issues.
I understand that the community is asking for this feature. As I always wrote this is a task for a monitoring software like nagios, ... To integrate it into xmr-stak direct the number of dependencies will increase by at least 3. I have some design in mind where the dependencies for xmr-stak will not be increased but you get temperatures. This is currently a lower priority task I need to invest more time into the GUI development.
Cool task, i'm too looking for best way. Not decided yet what to use - simple rrdtool, collectd or nagios, some one use more heavy like grafana.
I had been using Zabbix with external script to pickup stats from /api.json
and parse them into a format Zabbix consumes reasonably.
At least to track hashrates over time, and then same can be done with logging output from whatever SMI polling tool works OK with your drivers.
But after while I decided I don't really care about monitoring that closely, power bill minus normal usage divided by cashed out XMR is always close enough for hand grenades or horseshoes.
@Spudz76 whould you care to make a readme addition about best practices for temperature, vents and power monitoring while using xmr-stak ? It may eliminate new requests and a few bug reports due to incompatibility with monitoring software. I try to use a few options under win10 without success but do not feel that it is the place to complain for problems caused by other software packages.
I don't see how micromanaging these metrics gains any perceptible profit. If you have $2 profit per day on a rig, but you spent an hour per day worrying about it, then you lose $8 per day assuming you are a nasty self-boss and pay yourself $10/hr.
If you crash the driver poking pointless metrics, you'd lose even more.
I just check pool earnings and daily KWh used and literally don't check the actual rigs until one of those two things seems wrong. More of an overview approach than zoomed in so far I can count the hash per electron.
So I would be the wrong one to document how to do so, when I'm very anti-doing-so.
I'm not sure if this will be in Release compiles, but could be added and enabled on self-compiles Although there is a different way of loading the libraries at runtime and not linking to any of them so it might also be possible to have an option for it (and never touch the libraries if turned off, which is fine by me)
Do you have plan to add a option to report about: temperature, fan speed VGA, Mem,Core...? everytime I want to check temperature or fan speed VGA, I must open MSI afterburner. It is very inconvenience. Please add them into your tool.