Author

Topic: [Beta] Headless, centrally managed, turn key mining rig solution (Read 1233 times)

newbie
Activity: 31
Merit: 0
Made some big changes. Now rig configs are saved as SQL records instead of txt files. This lets me use a DEFAULTS template, then a rig's config would only have the settings that are different from the template. Also updated the Diablo miner and setup a system to easily update the miner on bootup. Also added more stats collection like accepts, rejects, hwerrors, and the reject ratio % of the rig.

Rigs can now easily be configured from a phone and reloaded or rebooted from a phone/browser.
member
Activity: 112
Merit: 10
Looks nice, very cool to see an IOS app Smiley up next, a wallet ? Wink
member
Activity: 84
Merit: 10
This account was a fraudulent account.
I would love to try this.
I'm currently running Ubuntu 11.04
legendary
Activity: 1428
Merit: 1000
https://www.bitworks.io
Great looking project!
newbie
Activity: 21
Merit: 0
Looks really cool. Can't wait to see the finished product.
newbie
Activity: 31
Merit: 0
LOVE IT! A question...

* GPU temps watched for over heating. When a temp gets over a configurable value per rig the miners are shut down and you are emailed.

I like the above option as a fail safe but I would also like an option to automatically adjust the fan speed in relationship to the GPU core temps. Keep below XX C type of thing with the software ramping up the speed 1% every 5 secs or so until point met. Can be done? Thanks for a cool utility!! Let me know when I can get my hands on it to check it out.

Ya that can be done but I'm not convinced it's worth the time. Running the GPU fan at 100% 24x7 is a good way to burn up the fan quickly. I'm keeping it at 85% so they'll last. That system would save a few watts but cost heat and for me the heat is more expensive to deal with then the kWh.
newbie
Activity: 40
Merit: 0
Not bad this could make it simple enough for the average user to join in.
newbie
Activity: 8
Merit: 0
LOVE IT! A question...

* GPU temps watched for over heating. When a temp gets over a configurable value per rig the miners are shut down and you are emailed.

I like the above option as a fail safe but I would also like an option to automatically adjust the fan speed in relationship to the GPU core temps. Keep below XX C type of thing with the software ramping up the speed 1% every 5 secs or so until point met. Can be done? Thanks for a cool utility!! Let me know when I can get my hands on it to check it out.
newbie
Activity: 31
Merit: 0
I just got some 6870's in so I was able to test things with SDK24 and 6000 cards. A few bugs had to be fix and now it's auto detecting SDK's and setting them up for you for both 5XXX and 6XXX cards.
hero member
Activity: 619
Merit: 500
Sounds very interesting.
Definitely something I might be interested in in the future.
newbie
Activity: 31
Merit: 0
I'm testing the waters to see if my creation has interest.

http://IOsMiner.com

http://iosminer.com/index.php/FEATURES
http://iosminer.com/index.php/IPhone_HUD

Features list copy:
    *  Ubunto 11.04 base system, USB thumb drive or SATA drive, 32bit for maximum compatibility.
    * OS optimized for headless operation. FSCK run automatically if need be with no prompting. RO / remount on errors.
    * iPhone optimized web page HUD that gives you a total cluster Mh/s and a quick view of the status of all your rigs, GPU temps (c/f), hash rates, last report in time, stale ratio %, hwerrors. IPhone_HUD
    * Headless remote configuration over HTTP based on eth MAC address. Much like VoIP phones.
    * GPU miner used is Diablo (+3% patched).
    * Some directories are moved over to a RAM drive to speed the rig up and save thumb drive write cycles.
    * Optionally set an expected GPU count setting on a rig. If the rig doesn't detect this many GPUs it emails you and reboots.
    * VoIP like central configuration. Configuration files are simple text files on a web server based on the rig's eth0 MAC address.
    * GPU count on rigs is fully automatically detected.
    * GPU mining Mh/s graphing.
    * GPU temp graphing.
    * GPU miner optimization switches adjustable per rig.
    * GPU miner is watched and if the miner fails it will automatically try to kick start the miner back up. If that works you are emailed about the incident and the Mhs is reported. If it is unable to kick start the miner automatically you are emailed and the rig is scheduled to reboot automatically in 1h if you don't cancel the scheduled shutdown to take a look at it yourself.
    * GPU core Mhz clocking. Automatically clock all GPU's the same speed or set each GPU to a different clock rate.
    * Also graph rig load, interface statistics, memory statistics, hard drive usage, etc.
    * Successful boot up message with additional rig information and debug logs to your email.
    * Xconfig automatically setup on boot so 1 OS image works on rigs with 1 card or 4.
    * GPU temps watched for over heating. When a temp gets over a configurable value per rig the miners are shut down and you are emailed.
    * Central RRD stats monitoring. Rsync back to your central server for graph collection.
    * Log cleanup and mail clean up on rigs.
    * Automatically logs into your private VPN network for secure remote TCP/IP access from anywhere through any firewall or NAT on the rigs LAN. (Hamachi)
    * SSH key management. Allows you to remotely access the server through SSH with no password. Instructions included. Great for scripts accessing the rigs through SSH.
    * Local rig /etc/hosts file is automatically updated with VPN network hostnames and private VPN addresses so all rigs can communicate over the VPN by hostname/nick-name alone. This means even when the rigs VPN IP address changes everything will be ok. Updated hourly automatically.
    * DHCP automatic IP assignment and local DNS server setup.
    * Can use either SDK 2.4 or SDK 2.1, automatically downloaded and installed if they aren't already setup.
    * Automatic time sync over NTP with pool.ntp.org.
    * SMTP relay server setting for email reports.

Added 6/15/11
* Automatically checks for not only a running GPU miner, but also that all GPU's on the rig have a load. If a 0% load is detected it attempts to self heal by restarting the miner. If the miner restart doesn't work it emails you and schedules a reboot in 5 minutes that you can SSH in and cancel if need be. Verified it works as designed on a dual GPU rig of mine I have this problem with sometimes.

Added 7/5/11
* Boot drive imaging script now works much better. It works on anything from 4gig USB sticks to 2TB SATA drives. Partitions are dynamically and automatically created and GRUB is installed and setup based on the drives UUID
* Rig configs moved from txt files to SQL records so we can use templates and default settings now.
* Rig configuration is done from a web page now instead of vi and a text file on a web server.
Jump to: