OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.10.0

[-ck]

Thanks. Core dump or debugging output is not much help without debugging symbols. Either build with the debug symbols (add -g) or grab a debug build from
http://ck.kolivas.org/apps/cgminer/debug/

[1498350032]

1498350032

[-ck]

Thank you, Kano and mdude77.

I know that failover-only would stop the leaking, however I've always used failover to get the highest efficiency, moreover when a new block is found and some pools are quicker than others.

The strange thing is that if I use failover-only, my Mh/s stay pretty the same in the middle of a block.

So even if cgminer thinks the primary pool is lagging (and so, with failover policy, it begins to collect a lot of work from elsewhere), this doesn't look to be true. If this were true, I think that with failover-only I should see my GPUs slow down from time to time, left with not enough work from my primary pool: but this doesn't happen.

It's not an exact science, and kicks in when it's below a threshold rather than waiting for it to fall to "not enough". It's an algorithm, and like all of them, it is prone to not being perfect. Local work generation via something like stratum makes this feature pretty much unnecessary.

[Lem]

It's not an exact science, and kicks in when it's below a threshold rather than waiting for it to fall to "not enough". It's an algorithm, and like all of them, it is prone to not being perfect.

I understand. I just say that this algorithm worked better (for me, eh) in previous version, before 2.9.x.

BTW: is there a way for me to modify this threshold, maybe working on some cgminer parameters (queue, scantime, expiry...)?

Local work generation via something like stratum makes this feature pretty much unnecessary.

Were it to me, stratum would become the default... yesterday.
But I think many little pools, and expecially the last "starving" prop pools - that are still there, though - won't rush to implement it. It's a pity.

[derjanb]

Thanks. Core dump or debugging output is not much help without debugging symbols. Either build with the debug symbols (add -g) or grab a debug build from
http://ck.kolivas.org/apps/cgminer/debug/

Done.

Code:

Reading symbols from /home/user/hdd/bitcoin/cgminer-2.9.4-x86_64-built/cgminer...done.
[New LWP 27740]
[New LWP 23114]
[New LWP 23132]
[New LWP 23099]
[New LWP 22458]
[New LWP 23138]
[New LWP 22756]
[New LWP 23135]
[New LWP 24650]
[New LWP 23134]
[New LWP 23133]
[New LWP 23113]
[New LWP 27743]
[New LWP 22757]
[New LWP 25467]
[New LWP 23100]
[New LWP 26811]

warning: Can't read pathname for load map: Input/output error.
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Core was generated by `./cgminer -c ../cgminer.conf --gpu-engine 500-600 --temp-cutoff 85 --gpu-fan 45'.
Program terminated with signal 6, Aborted.
#0  0x00007fa5177ac425 in __GI_raise (sig=<optimized out>) at ../nptl/sysdeps/unix/sysv/linux/raise.c:64
64      ../nptl/sysdeps/unix/sysv/linux/raise.c: No such file or directory.
(gdb) bt
#0  0x00007fa5177ac425 in __GI_raise (sig=<optimized out>) at ../nptl/sysdeps/unix/sysv/linux/raise.c:64
#1  0x00007fa5177afb8b in __GI_abort () at abort.c:91
#2  0x00007fa5177ea39e in __libc_message (do_abort=2, fmt=0x7fa5178f1e3f "*** %s ***: %s terminated\n") at ../sysdeps/unix/sysv/linux/libc_fatal.c:201
#3  0x00007fa517880807 in __GI___fortify_fail (msg=0x7fa5178f1dd6 "buffer overflow detected") at fortify_fail.c:32
#4  0x00007fa51787f700 in __GI___chk_fail () at chk_fail.c:29
#5  0x00007fa51787eb69 in _IO_str_chk_overflow (fp=<optimized out>, c=<optimized out>) at vsprintf_chk.c:35
#6  0x00007fa5177f213d in _IO_default_xsputn (f=0x7fa4f4ff8ba0, data=<optimized out>, n=2906) at genops.c:485
#7  0x00007fa5177c0f82 in _IO_vfprintf_internal (s=<optimized out>, format=<optimized out>, ap=<optimized out>) at vfprintf.c:1630
#8  0x00007fa51787ec04 in ___vsprintf_chk (s=0x7fa4cc0020e9 "0100000001", '0' <repeats 64 times>, "ffffffff4c03d92e030f00456c69676975730050aa1ca20488fabe6d6d5ac523f49912c85158966f09972f3689830ac7371984a4d0e71335bb878f56b90800"..., flags=1,
    slen=512, format=0x448ff2 "%s", args=0x7fa4f4ff8cc8) at vsprintf_chk.c:86
#9  0x00007fa51787eb4d in ___sprintf_chk (s=<optimized out>, flags=<optimized out>, slen=<optimized out>, format=<optimized out>) at sprintf_chk.c:33
#10 0x000000000040fc75 in sprintf (__fmt=0x448ff2 "%s", __s=0x7fa4cc0020e9 "0100000001", '0' <repeats 64 times>, "ffffffff4c03d92e030f00456c69676975730050aa1ca20488fabe6d6d5ac523f49912c85158966f09972f3689830ac7371984a4d0e71335bb878f56b90800"...)
    at /usr/include/x86_64-linux-gnu/bits/stdio2.h:34
#11 gen_gbt_work (pool=0x10ec310, work=0x7fa4cc001e40) at cgminer.c:1487
#12 0x0000000000414e0b in get_work_thread (userdata=0x7fa50000a770) at cgminer.c:3066
#13 0x00007fa5188a2e9a in start_thread (arg=0x7fa4f4ff9700) at pthread_create.c:308
#14 0x00007fa517869cbd in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:112
#15 0x0000000000000000 in ?? ()

[gyverlb]

First, thanks for your hard work!

I've updated my Ubuntu 12.04 x64 today and since the reboot cgminer 2.9.* creates coredumps. Version 2.8.7 is working fine, 2.9.3 and 2.9.4 are not. The core dump is created randomly after 1 to 30 seconds after cgminer was started.

This is the gdb output:

May be the same problem than the one I reported on IRC (and started for me at 8AM GMT today), if it is, adding --fix-protocol should be a temporary fix (it disables the switch to GBT protocol).

[Subo1977]

The 2.9.3 runs the last day's without problems.

since this morning i have a problem with cgminer 2.9.3 and 2.9.4 on Windows and Linux.

on these pool's its ok. : eclipsemc, and Eligius.
after switching  or starting to p2pool the cgminer crashs after a few seconds.

after reverting to 2.8.7 already is ok.

[derjanb]

First, thanks for your hard work!

I've updated my Ubuntu 12.04 x64 today and since the reboot cgminer 2.9.* creates coredumps. Version 2.8.7 is working fine, 2.9.3 and 2.9.4 are not. The core dump is created randomly after 1 to 30 seconds after cgminer was started.

This is the gdb output:

May be the same problem than the one I reported on IRC (and started for me at 8AM GMT today), if it is, adding --fix-protocol should be a temporary fix (it disables the switch to GBT protocol).

You're right "--fix-protocol" fixes cgminer too. Funny thing, cause 2.9.3 was working fine the last 3 days.

[gyverlb]

You're right "--fix-protocol" fixes cgminer too. Funny thing, cause 2.9.3 was working fine the last 3 days.

I use p2pool, but apparently Bitminter and Eligius (which both use GBT) in my backup pools can trigger the bug since this morning even if they aren't used.
Posted a backtrace to #cgminer, hopefully it will help debug the problem.

[rav3n_pl]

You're right "--fix-protocol" fixes cgminer too. Funny thing, cause 2.9.3 was working fine the last 3 days.

I use p2pool, but apparently Bitminter and Eligius (which both use GBT) in my backup pools can trigger the bug since this morning even if they aren't used.
Posted a backtrace to #cgminer, hopefully it will help debug the problem.

If you mine in p2pool use another p2pool node as backup. There is few public nodes open for everyone.

[Luke-Jr]

You're right "--fix-protocol" fixes cgminer too. Funny thing, cause 2.9.3 was working fine the last 3 days.

I use p2pool, but apparently Bitminter and Eligius (which both use GBT) in my backup pools can trigger the bug since this morning even if they aren't used.
Posted a backtrace to #cgminer, hopefully it will help debug the problem.

If you mine in p2pool use another p2pool node as backup. There is few public nodes open for everyone.

None of those (AFAIK) support decentralized mining, though. Nor is it really practical as GBT's bandwidth use is absurd with p2pool's 10 second blocks.
There's no reason to prefer p2pool over Eligius or Bitminter anyway, so long as you're decentralized mining.

[kano]

You're right "--fix-protocol" fixes cgminer too. Funny thing, cause 2.9.3 was working fine the last 3 days.

I use p2pool, but apparently Bitminter and Eligius (which both use GBT) in my backup pools can trigger the bug since this morning even if they aren't used.
Posted a backtrace to #cgminer, hopefully it will help debug the problem.

If you mine in p2pool use another p2pool node as backup. There is few public nodes open for everyone.

None of those (AFAIK) support decentralized mining, though. Nor is it really practical as GBT's bandwidth use is absurd with p2pool's 10 second blocks.
There's no reason to prefer p2pool over Eligius or Bitminter anyway, so long as you're decentralized mining.

FFSake can you stop with the FUD - Eligius and Bitminter are not decentralized - learn English, not whatever warped, deluded, zealot language you use.

There is a perfectly simple reason to use p2pool and not go anywhere near ANY centralised GBT pool like Eligius or Bitminter and that is the 2 orders of magnitude larger amount of data GBT transfers across the internet to your miner compared to the old getwork or new stratum pool protocol.
Even a bitcoin dev has said that the GBT data design is not designed for a WAN (or even a LAN)

The RPC protocol was never designed to minimize data transfer. It's meant for controlling a bitcoind instance on the localhost, in which case the amount of data is of no essence (hence the choice of JSON, too).

... as per my Issue I raised on the bitcoin code (where you sneaked in your botnet support into bitcoind/GBT)
https://github.com/bitcoin/bitcoin/issues/1985
I also do not understand why you supported botnets on Eligius and now added support of them into bitcoind that everyone now has in their bitcoind

Bitminter and Eligius ARE centralised - there is one place that controls all the work - the pool - just like every other non-p2pool pool does.
On Eligius GBT, the centralised pool decides what is valid work and what is not valid work.
On p2pool it is the p2peers that decide ... and if suddenly half the peers decide one type of work is valid and the other half decide a different type of work is valid, you will then get 2 p2pools running with 2 different sharechains
Be it half the miners or just one miner, the miner can keep mining no matter what it decides is valid - each sharechain is simply the list of p2pool miners that agree in it's contents ... and that is by definition decentralised.

[Roy Badami]

It looks like stratum backup pools aren't detected as alive when specified with a stratum+tcp:// URL (though, strangely, they work fine if you put in an http:// URL pointing to the stratum port).

At least with EMC.

ETA, unless I don't understand what a stratum URL is supposed to look like.

With the default (failover) strategy, if I specify my pools as follows then only whichever pool I put first is detected as alive (and if that's down it *doesn't* failover).

stratum+tcp://us1.eclipsemc.com:3333
stratum+tcp://us2.eclipsemc.com:3333
stratum+tcp://us3.eclipsemc.com:3333

If I specify them as follows then it connects with stratum, detects all pools live, and failover seems to work:

http://us1.eclipsemc.com:3333
http://us2.eclipsemc.com:3333
http://us3.eclipsemc.com:3333

ETA again: Although to be fair some of the evidence that lead me to the above conclusion was collected from a version of git master HEAD somewhere between 2.9.3 and 2.9.4.  But I'm certainly seeing backup pools report as down in 2.9.4 - can't say for sure that failover is actually not working.

[mdude77]

alas, 2.9.4 crashed on me today on my windows 8 box while pointing to oz.  11 hours before I got home from my day job.

I'm going to try --verbose and see if it stops doing that.

M

[-ck]

alas, 2.9.4 crashed on me today on my windows 8 box while pointing to oz.  11 hours before I got home from my day job.

I'm going to try --verbose and see if it stops doing that.

M

Did you have a GBT pool as a backup somewhere? I think there's a crash unique to pools that send GBT info. If so, --fix-protocol should prevent it, so you'll have to specify the stratum pools' urls/ports directly.

[mdude77]

alas, 2.9.4 crashed on me today on my windows 8 box while pointing to oz.  11 hours before I got home from my day job.

I'm going to try --verbose and see if it stops doing that.

M

Did you have a GBT pool as a backup somewhere? I think there's a crash unique to pools that send GBT info. If so, --fix-protocol should prevent it, so you'll have to specify the stratum pools' urls/ports directly.

I probably do.  I don't know which ones use GBT and which ones don't.

M

[louisBSAS]

Quick question @ckolivas

Just flashed over to check that all miner are up - saw this...




What is "Harmful miner version detect" - using 2.6.1 on bitminter?

I run Linux 10.10, don't have any backup pools, don't usually upgrade anything, because my rigs just mine continuously with no problems.

[juhakall]

What is "Harmful miner version detect" - using 2.6.1 on bitminter?

If you run a cgminer or bfgminer version below 2.7:
You are causing yourself and all other miners to lose transaction fee income on the blocks you make. You were causing slow server response and a high reject ratio, before these changes. You are getting work slower than others on the pool. You are probably getting more stales too. I know this isn't something you intended to do and you may not have been aware of it before. But you are running a very old miner version with some serious bugs in it. Upgrade today!

[louisBSAS]

... But you are running a very old miner version with some serious bugs in it. Upgrade today!

Thanks for the info.  That machine is just an old MB with a card plugged in. I guess I need to drag a monitor over to it and upgrade it.  My other rigs are running 2.7.6 with no problems. My GPU days are numbered anyway...

[DoomDumas]

I've searched the forum and this thread and could not find an answer..

Settings/Write Config ENTER to default cgminer.conf did not save engine and memory clock settings.

Each time I restart CGMINER I need to set memory and GPU clocks.

I'm using an old version (2.7.5) maybe this is an issue that have already been adressed.
I've also tried to edit manualy the cgminer.conf file, not effective.

Sorry for maybe noob question, I've search for an answer before asking !

Thanks anyone for your time.

[-ck]

I've searched the forum and this thread and could not find an answer..

Settings/Write Config ENTER to default cgminer.conf did not save engine and memory clock settings.

Each time I restart CGMINER I need to set memory and GPU clocks.

I'm using an old version (2.7.5) maybe this is an issue that have already been adressed.
I've also tried to edit manualy the cgminer.conf file, not effective.

Sorry for maybe noob question, I've search for an answer before asking !

Thanks anyone for your time.

It saves command line parameters, not things you change later on.