Topic: Avalon 741 needed rebooted (Read 1034 times)

Just as a data point:    This was happening right about every 10 days.   I then adopted a "reboot every Sunday" strategy.   Three reboots in three weeks and the problem has never shown up again.

So this points a guilty finger at software.   Probably nothing I can do about it but I will mention that my Avalon controller is one version behind.   The change log for the new version does't mention any stability fixes but is anyone aware of possible remediation in the latest version for this type of behavior?

Thanks in advance,

https://bitcointalk.org/Themes/custom1/images/smflogo.gif

Thanks folks!

Set-up
===========================
120V/60Hz/20Amp circuit and a GFCI outlet.
Nothing but my mining rig on circuit
2650W surge suppressor.  Everything is plugged into the suppressor
One Avalon 741.
AUC3 and Avalon controller.  
Standard USB wall wart, to power the controller.
Two Power supplies :  EVGA 650W SuperNovaG2 and also a 750W version.
    The 650 has two cables, both have two connectors.   So one side of the miner is powered with 3 of the four connectors.
    The 750W has 4 cables, so three of them power the other side.
AUC3 connected to controller and miner with cables supplied by Canaan

And the result of um, monster PSU's feeding un-fused 'measly' 16ga wire triplets... This s7 belonged to Tupsu as I recall Finsky, Those IBM 4kw supplies do make a great arc welder eh? The thread with several more pretty horrific pics of the conflagration is here https://bitcointalksearch.org/topic/m.15478853

That said, really do need a 101 here to consolidate the info.

Definitely a great idea, I've been reading old threads trying to build my troubleshooting knowledge base.

Mix that in with tales of random fire and majik smoke being released I may not wake up drilling next to my keyboard lol.

If you do fire one up, I've been looking for info on what all the lines mean in the advanced miner info... and navigating the kernel log info. I'm a sucker for punishment and never could just live in ignorant bliss lol

Yeah, I've done threads like that on my "Hacking xxxx miners for the hell of it threads". Given that people are sending Avalons maybe I should make a thread for it. Or if a mod is ok I could curate a master sticky thread of "So what the heck is my miner doing" or something like that.....

Fun stuff like what do the lights mean, why is my miner shooting flames out the end (oh wait, that's Titans), how do I plug it in, why are my plugs hot, what's a good running temperature, how do I get rid of heat (open window), why should I clean my miner (aforementioned fires caused by dust bunnies igniting), stuff like that.

Right now for example I'm running an A6 on a 750 watt corsair. It's actually quite quiet with the fan purring at 4260rpm, 2.4th, 30c, with 12v voltage at a nice relaxing 11.5 volts. Less than full blast if I put a pair of Corsairs on it, but still nice. I should hook up the E-meter to see what the draw at the wall is.

C

Ya know, anyone ever make a Basic trouble shooting 101 thread to cover the plethora of these same power-related and cooling queries that relate to all ASIC-based miners?

For things like #1: If using an ATX PSU always check its detailed specs to see if it has 1 supply rail or more than 1. Followed of course by 1a: If it has more than 1 rail NEVER power 1 card with more than just 1 rail. Joining rails together results in things a-poppin' and Majik Smoke being released...

Of course newbs will still blissfully ignore said stickied thread and post anyway but at least it would make a quick thread to link them to...

Hey there it may be that the brick powering the controller doesn't have enough power. What are you using?

How is your whole setup wired, including the miner power supplies and what circuits are everything on. I know you mention using a surge protector, what all is plugged into it?

Hopefully we can get this sorted out.

Power supply issue. What type of supplies are you using and what is the voltage at the miner itself?

OK - it happened again.   To recap:   The symptoms are: No work being sent to the pool, the controller unresponsive, no fans, but all lights are fine.

Miners:   Lights were perfect.  All green on all power supply inputs and green on main light
AUC3:  Lights perfect.   A green and a blue like normal.
Controller:  The controller has no loghts at all so nothing to see here but web interface was not responding.  Lan port flashed with activity like normal so not a power issue.

Rebooted the controller only, and system came back.   Ran fine for 40-ish minutes.   Left for work.  Then did the same thing.    Unfortunately this time I was not there and I just had my wife hit the power reset button on the power strip.  That was 3.5 hours ago and all is still well.


So my thought is there is something - something odd that only occurs with lots of work and uptime - that the controller is seeing from the miner or AUC3 that sends the controller into oblivion.


But how to check?  Logs are gone when the controller is non-responsive and you have power reset.    Any idea???

View numbers.  Joules says nothing about power consumption of a PSU.  Joules measures energy absorbed from a surge.  Potentially destructive surges are hundreds of thousands of joules.  How many joules will those older and new power strips absorb?

Your miner would easily convert a hundreds or thousand joule transient into rock stable, low DC voltages to safely power its semiconductors.  Transients that tiny are already made irrelevant by what is already inside electronics.  Your concern is a transient that is many times larger.  How many joules with that power strip absorb?

The most critical item on a power strip is its circuit breaker.  How many amps is that rated to provide?  How many amps does each miner demand?  Any decision without first learning these numbers is wild speculation resulting in a junk science conclusion.  What are the relevant numbers?

If a miner needs protection, then so do dishwasher, GFCIs, furnance, clocks, garage door opener, central air, vacuum cleaner, LED & CFL bulbs, refrigerator, dimmer switches, and everything else.  Protection (that safely absorbs hundreds of thousands of joules and remains functional for decades) should be considered.  This 'whole house' solution costs about $1 per protected appliance.

Only lights that say anything useful about AC power glitches are incandescent bulbs.  Power glitches can occur without lights on electronics reporting the anomaly.  Only a UPS would avert that anomaly.  Power strip protectors would do nothing.  Again, spec numbers define that.

Thanks.  I did have to buy a new surge protector since none of my spares would handle it.   So I bought one specifically for this outfit.  Don't remember the joules rating but I had plenty of headroom.   The air is deflected away from the power supplies, so I have that covered.  The power strip's lights were normal - I did notice that.   The controller definitely did not respond.   It is almost like it locked up, and the miner shut itself down based on the lack of communication.  I am supposing that because it was a software issue;  the lights on the miner (the two I could see casually without climbing) that indicate PSU power would remain on.

If it happens again, I'm pretty sure your lights will stay on for the Auc converter and the controller if it has lights.

The only lights that would go off due to a Power supply issue are on the Miners themselves. When I was first setting up my new Avalons I had to change around how I was directing air as my Power supply shut off after a few days due to a high Temp switch on the PSU itself. After that adjustment I had no further issues.

I am wondering when you said you have no response from the controller. Do you have your Miner and controller running of a Surge protector strip? If so you may want to check that it is rated to handle the amps for your miner? Good luck.

  I should have checked the lights closer.  I did have lights on the miner and did not notice anything unusual - just went straight for the power reset button on the power strip.  Did not check the AUC/controller.  It's all on a high shelf and while I can see the lights on the miner I can't on anything else without climbing and I just did not bother.

If you the avalon lost power and had no lights or fan that is 100% a power supply issue.

 I got a new 741 miner from Avalon Wednesday, had it up and running by Wednesday night. Everything's been going great, and I have been pleased. However sometime between 6 PM and 10 PM tonight  it went black and needed to be rebooted. No fans, and no response from the controller. After rebooting everything was fine. Have other people experienced less than one week uptime with these units?

 It did coincide with some local stormy weather, but the weather had passed before this occurred and I was here the whole time and never saw any indication of a black/brown out on any of my other equipment or lights.

PS I have the equipment on a surge protector, but I do not have it on a UPS.