Mining setup - Hard reset / uptime management/ remote options.

[gox1201]

Greetings.

Just a post to try get some ideas on hardware management options. To clarify, this is not about setting up a miner for maximal hash rates etc, but rather about the things you do to make it run smoothly, without you baby sitting it 24/7.

I'm assuming everyone does the minimum stable drivers, no HW error mining etc.

So I'll start off with a few statements of what I've been doing, then ask a few questions and hopefully we can get some decent input that way.


My methods to insure close to maximum uptime at full mining rate

1. CGWatcher & CGRemote
- Beautiful little programs, allows me to have it automatically detect sick or dead GPUs, and I've set it so that if my mining hashrate falls under 850 for my 290 cards (which sit at 900-905), that it will restart CG Miner, it can also do the same for a sick/dead card.

2. Windows/CGWatcher
- My system seems stable for indefinite periods, but I'm away from home for >12h stints at a time, and if I was to have downtime in that period, I would lose all income until such a time as I'm back.
- So as a tradeoff, I've setup windows and cgwatcher to do alternating 6h reboots (aka windows reboots every 12h, and cgwatcher reboots every 12hours, but staggered so its every 6 hours in truth)
- This allows sufficient failover incase e.g. cgwatcher gets stuck in a no-api interaction error, where it essentially loses control of the system.

A reboot for my system, essentially means less than a minute that I'm not mining, so 1min every 6 hours downtime is vastly preferential for me to say having 5-8hours of downtime say once a week while at work.


Now my questions:

1. I had an issue on Friday around midnight, where my DSL had picked up a glitch, and essentially was a dead connection. Thus not able to connect to my mining pool, thus no income. My normal failovers of course didn't pick this up, and as such I lost a couple of hundred rand of income.

- Does anyone have a nice way of monitoring their router / forcing it to reboot say once every 24 hours or whatever?

2. I have not encountered a scenario yet where my hardware stalls out completely, but it will become more frequent once I start pushing very close to the max of what my boards could support.

- Does anyone have a nice way for hard resetting their plug / wall switch / pre-PSU area?

The only thing I've managed to think of so far, would be a plug-timer, but I don't like the idea of generating a powerloss just to reboot the system randomly, especially if it was happily mining and thus not doing a decent "shut-down" sequence.

- Ideally the system I would want would need to be able to be remote (aka over IP from say a cellphone) controlled, to effect a "reset" on the plug?


If anyone has any input, it would be greatly appreciated. Also if you have any issues, ask away and perhaps we can come to some sort of answer, so that everyone doesn't have to re-invent the wheel.

All the best
me

 

[SoloMid]

I cant add much input here bud. Im not half as organised as you. Only suggestion I could try make is with regards to the router. Get a Dsl/3G router with automatic failover thus ensuring a constant connection.

You can also get apps for android which allow you to monitor your hashrate, so you can see if you are still digging away. Obviously Teamviewer mobile too.

Well thought out tho buddy

 

[Pansyfaust]

I'm having a *** time setting up my miner in my office at uni. Twice now I have have been there and the setup runs fine, literally as soon as I am home the miners stop working. I set it up so that if there is a power outage that as soon as the power comes on the PC autoboots, loads up cgminer, automatically loads up msi afterburner settings and connects to the internet. So I have no idea why two nights in a row the PC is not hashing.

I also cant use any sort of remote desktop because I have no idea why TightVNC cant connect to the PC. Anybody have any experience trying to do remote desktop over a 3g dongle? Anybody used Chrome Remote Desktop?

Also having another strange issue that must be driver related.

I have a 50/50 chance that whenever my PC restarts and cgminer autostarts one gpu core will not be found. I get something like this:

ADL found less devices than opencl!
There is possibly more than one display attached to a GPU
Use the gpu map feature to reliably map OpenCL to ADL
WARNING: Number of OpenCL and ADL devices did not match!
Hardware monitoring may NOT match up with devices!"

Is this a SDK problem? As far as I've read amd's 13.12 drivers comes with the proper SDK already. Never had this issue when I was mining on my main PC, but now that its moved to my dedicated mining PC I am having issues with it. And like I said, its so random. One restart, everything is golden. Next restart, no dice.

 

[Pansyfaust]

Also, not sure if you guys noticed, but closing any HW monitoring apps increasing your hash rate. Not sure if this applies if you are running your desktop off onboard, but I tried setting it up that way and cgminer had a hissyfit, so just defaulted back to normal.

 

[abs1]

Greetings.

Just a post to try get some ideas on hardware management options. To clarify, this is not about setting up a miner for maximal hash rates etc, but rather about the things you do to make it run smoothly, without you baby sitting it 24/7.

I'm assuming everyone does the minimum stable drivers, no HW error mining etc.

So I'll start off with a few statements of what I've been doing, then ask a few questions and hopefully we can get some decent input that way.


My methods to insure close to maximum uptime at full mining rate

1. CGWatcher & CGRemote
- Beautiful little programs, allows me to have it automatically detect sick or dead GPUs, and I've set it so that if my mining hashrate falls under 850 for my 290 cards (which sit at 900-905), that it will restart CG Miner, it can also do the same for a sick/dead card.

2. Windows/CGWatcher
- My system seems stable for indefinite periods, but I'm away from home for >12h stints at a time, and if I was to have downtime in that period, I would lose all income until such a time as I'm back.
- So as a tradeoff, I've setup windows and cgwatcher to do alternating 6h reboots (aka windows reboots every 12h, and cgwatcher reboots every 12hours, but staggered so its every 6 hours in truth)
- This allows sufficient failover incase e.g. cgwatcher gets stuck in a no-api interaction error, where it essentially loses control of the system.

A reboot for my system, essentially means less than a minute that I'm not mining, so 1min every 6 hours downtime is vastly preferential for me to say having 5-8hours of downtime say once a week while at work.


Now my questions:

1. I had an issue on Friday around midnight, where my DSL had picked up a glitch, and essentially was a dead connection. Thus not able to connect to my mining pool, thus no income. My normal failovers of course didn't pick this up, and as such I lost a couple of hundred rand of income.

- Does anyone have a nice way of monitoring their router / forcing it to reboot say once every 24 hours or whatever?

2. I have not encountered a scenario yet where my hardware stalls out completely, but it will become more frequent once I start pushing very close to the max of what my boards could support.

- Does anyone have a nice way for hard resetting their plug / wall switch / pre-PSU area?

The only thing I've managed to think of so far, would be a plug-timer, but I don't like the idea of generating a powerloss just to reboot the system randomly, especially if it was happily mining and thus not doing a decent "shut-down" sequence.

- Ideally the system I would want would need to be able to be remote (aka over IP from say a cellphone) controlled, to effect a "reset" on the plug?


If anyone has any input, it would be greatly appreciated. Also if you have any issues, ask away and perhaps we can come to some sort of answer, so that everyone doesn't have to re-invent the wheel.

All the best
me

My suggestion would be a remote relay over gsm connection, some comes with fancy apps you can run on your pc, most home automation companies make use of these devices.

7 Switch GSM Remote Control Relay Quad Band | eBay

Donations are welcomed..lol

LTC: LSBuckTPbPGtKxoW4hKDaGpFAf43T3Hmfd

 

[abs1]

Also, not sure if you guys noticed, but closing any HW monitoring apps increasing your hash rate. Not sure if this applies if you are running your desktop off onboard, but I tried setting it up that way and cgminer had a hissyfit, so just defaulted back to normal.

Some people do say so but i see the benefits in management apps, a couple of hash wont hurt if you think of the benefits it offers, if you clever enough you can use the miners api and report only what you want like temps and so forth.

 

[PsyWulf]

1. I had a Cron/Win scheduler job to telnet into my routers and reboot them once every 24hours. My reply on point2 might be useful too in this regard if the router software is unresponsive
2. With some Ubiquity mFi plugs you should be able to remotely "force" power cycle any device connected to the plug
mFi | Ubiquiti Networks, Inc.
mFi 3-port Power with WiFi

Can set up rules and schedules using mFi so might be doable to have a timed restart to the plug connected to the router for shorter cycling,and a longer cycling period or maybe a cycle if the system stops responding to ping requests or similar

 

[roligov]

Think I mentioned it to you already, but I'm using a rasp Pi and have hooked up the GPIO pins to a 8 way relay board which in turn is hooked up to the power switch pins on each mobo. Then I have a Web interface that is hosted by Apache on the raspi and it has on and off buttons for each rig that calls python commands to power on and off rigs individually. Helps a lot if you need to hard power off a rig and are at work.

Sent from my GT-N7100 using Tapatalk

 

[Oj0]

I've got remote desktop working over pretty much any connection, including my phone using HSDPA or WiFi. I'm using Team Viewer.

 

[gox1201]

I had another bugger up today, literally 10 mins after leaving for work my second miner lost internet connectivity.

Its got an ethernet cable directly into my router.

Router kept working, but is somehow getting "corrupted" and can only be fixed by powering the router off and back on
- essentially my wireless devices say the signal from the router is strong, but that it has no internet connectivity
- some of the pc's happily run off it in this state still, with the internet working
- and some say "unknown connection" "limited connectivity"

All this does do however, is make CGminer give an "unknown url" error, which effectively buggers up mining until I reset the router.

REALLY frustrating

 

[Pansyfaust]

Another night and the PC at Uni is not mining :/. The problem last night was the cgminer crashed. Today I set up CGwatcher to restart the PC if no shares change in 10 minutes, but still something has farked it up :/ and obviously it is not restarting.


Think I must just return the PC home to mine, its becoming tiresome.

 

[Lederstrumpf]

TeamViewer has been working very well for me so far. Not the most secure way to do it, but I dump my wallets to cold storage once they reach a certain threshold either way. I'll gladly risk losing a fraction of them for the sake of the coupled convenience - ATI driver hassles are already eating up enough of my time as it is.

 

[eezolive]

I looked into this some time back aswell, the cheapest solution i found was a hack job on the serial port to the reset pin of the mobo, so you have to daisy chain them, 1 pc can switch of the next one

I'm trying to find the link of how to use the serial port to short out the reset pin

 

[Pansyfaust]

I had another bugger up today, literally 10 mins after leaving for work my second miner lost internet connectivity.

Its got an ethernet cable directly into my router.

Router kept working, but is somehow getting "corrupted" and can only be fixed by powering the router off and back on
- essentially my wireless devices say the signal from the router is strong, but that it has no internet connectivity
- some of the pc's happily run off it in this state still, with the internet working
- and some say "unknown connection" "limited connectivity"

All this does do however, is make CGminer give an "unknown url" error, which effectively buggers up mining until I reset the router.

REALLY frustrating

Now I am dealing with this as well. Sometimes the Dongle also looses connection, and cgminer just sits idle and CGWatcher does not mind this :/. Also lost about 4 hours last night coz the dongle tripped.

If you find a solution make me aware please.

 

[gox1201]

Seems to be a dhcp error, billion router. Haven't found a solution yet

Sent from my GT-N7100 using Tapatalk

 

[Senshi]

Seems to be a dhcp error, billion router. Haven't found a solution yet

Sent from my GT-N7100 using Tapatalk

Run static IP's and disable DHCP.

 

[Alien]

create a remote log in to the router? both my routers support remote login.that way you can reset it remotely. or if you are on static IP...maybe even reset just the one pc

 

[Senshi]

create a remote log in to the router? both my routers support remote login.that way you can reset it remotely. or if you are on static IP...maybe even reset just the one pc

And if the router becomes non-responsive?

 

[Alien]

And if the router becomes non-responsive?

BUILD A ROBOT WITH A PAPERCLIP IN ITS HAND
If you have a parallel port and 3g dongle then it can be done remotely...maybe even automatically. ping google fails-> batch runs 3g dongle which notifies you(if your pool doesnt already) then you log in and send a signal(5v i think) over the parallel port which would need to be hardwired to the router's reset.

Of course this is theory and assuming that you can work out the routers circuitry