ethminer 0.16.0.dev1 can't start miner - Error: "terminate called without an active exception" (smOS)

[luteknarozniak]

Getting error message:

============Starting Miner ===============
m 18:54:05 ethminer ethminer 0.16.0.dev1
m 18:54:05 ethminer Build: linux/release
terminate called without an active exception
/root/xminer.sh: line 102: 4666 Segmentation fault $MINER_PATH $MINER_OPTIONS
Miner ended/crashed. Restarting miner in 10 seconds --------------

Tried with nanopool, dwarfpool, sparkpool and nicehash - makes no difference.
Tried with overclock applied and stock values as well - again, no difference.

No problems with ethminer 0.14 (there's no 0.15 available on smOS to try).

Developer of smOS claims they have tested it on nanopool and it worked. Suggested to ask around on ethminer forum.

Any ideas?

Lutek.

[AndreaLanfranchi]

Is it the first start or are you re-starting after an error ?

Could you please do a cold restart and start the miner ? (report results)

[luteknarozniak]

Error occurs every time I'm switching miners from v.0.14 to v.0.16.dev1. Doesn't matter whether I'm just reloading miner or restarting the whole rig.
Will do a cold restart next.

[luteknarozniak]

Cold restart done. Same error:

terminate called without an active exception
/root/xminer.sh: line 102: 4030 Aborted $MINER_PATH $MINER_OPTIONS
Miner ended/crashed. Restarting miner in 10 seconds --------------
============ Starting Miner ===============
m 20:35:44 ethminer ethminer 0.16.0.dev1
m 20:35:44 ethminer Build: linux/release
terminate called without an active exception
/root/xminer.sh: line 102: 4822 Aborted $MINER_PATH $MINER_OPTIONS
Miner ended/crashed. Restarting miner in 10 seconds --------------
============ Starting Miner ===============
m 20:35:54 ethminer ethminer 0.16.0.dev1
m 20:35:54 ethminer Build: linux/release
terminate called without an active exception
/root/xminer.sh: line 102: 5508 Aborted $MINER_PATH $MINER_OPTIONS
Miner ended/crashed. Restarting miner in 10 seconds --------------

[smurfy]

what exactly is the commandline?

[luteknarozniak]

-U -P stratum1+tcp://ETH_address.Worker_name@eth-us-west1.nanopool.org:9999

also tried:
-U -P stratum2+tcp://BTC_address.Worker_name@daggerhashimoto.usa.nicehash.com:3353 (for nicehash)
-U -P stratum1+tcp://ETH_address.Worker_name@eth-us2.dwarfpool.com:8008 (for dwarfpool)

etc.

[ddobreff]

We have switched to autostratum configuration.

[luteknarozniak]

We have switched to autostratum configuration.

Is my command line wrong then?

[xdemocle]

I don't think is command line misconfiguration.
I got same issue with my rig nvidia 1060:
terminate called without an active exception

[StefanOberhumer]

@xdemocle : Are you also using SMOS (Simple Mining OS) ?

[xdemocle]

@StefanOberhumer no sorry, forgot to mention. I do not use SMOS.

[StefanOberhumer]

OK - I can reproduce it if I use the official 0.16.0.dev1 release (normally I use self compiled actual dev snapshots).
Seems to be a build problem of the official release !

[luteknarozniak]

Perhaps Tytanick (smOS developer) had tested the miner compiled from the same source as you @StefanOberhumer and it worked for him, but then he has added the official release to smOS and that one, unfortunately, comes with some build issues.

[StefanOberhumer]

@chfast : As I don't know where and how the releases are built - can you help or supply info how to build an official release (which compilers,.... - or is it built via Travis/CircleCI/AppVeyor) ?

[XhmikosR]

@StefanOberhumer It's built on Travis for nix and AppVeyor for Windows.

[StefanOberhumer]

@xdemocle , @luteknarozniak :
Can you try ./ethminer -U --list-devices
This results in my case in X 09:16:33 ethminer CUDA error: Insufficient CUDA driver: 9010 which is correct for my environment.
Have to check where ethminer segfaults if I just start mining.

[xdemocle]

error: Insufficient CUDA driver: 9010 could be the reason. That would explain why on my other nvidia rig with latest cuda drivers ethminer is able to run.
But I would have time to try on this rig with 1060 GPUs only by tonight. Sorry

[ddobreff]

This was already discussed. CUDA 9.2 works on 396+ but not on 390, while 9.1/9.0 work on both. There is no known/found benefit of using 9.2 over 9.1/9.0.

[StefanOberhumer]

@ddobreff : I think it's not an issue/discussion which CUDA driver version we're using/require.
Currently ethminer segfaults without displaying a proper message if the driver mismatches - That should be fixed.

[StefanOberhumer]

@luteknarozniak : Just as I know/remember:
smOS does not upgrade the drivers/kernel within their online updates!
To get new drivers/kernel you have to download their latest image and write it to the HDD/USB!

[luteknarozniak]

@luteknarozniak : Just as I know/remember:
smOS does not upgrade the drivers/kernel within their online updates!
To get new drivers/kernel you have to download their latest image and write it to the HDD/USB!

@StefanOberhumer I flushed new image (kernel 4.7.1) about couple of weeks ago. I don't think there is a newer version.
Just run nvidia-smi command and yes, we're still running on Driver Version: 390.67.
Will ask Tytanick if he can install 396 driver for us.

[ddobreff]

396 are not stable enough to use in production environment.

[StefanOberhumer]

@luteknarozniak
1.) Do you get also error: Insufficient CUDA driver when running ./ethminer -U --list-devices
==> we can close this issue
2.) You can add some comments to issue #1390 where a downgrade to CUDA 9.0 is requested

[luteknarozniak]

@StefanOberhumer

1.) Do you get also error: Insufficient CUDA driver when running ./ethminer -U --list-devices
==> we can close this issue

Can't run that command as there is no ./ethminer folder in smOS (at least I can't find it). However, I'm getting Insufficient CUDA driver: 9010 error while trying to mine with ethminer 0.15.0, Tytanick has just added.

2.) You can add some comments to issue #1390 where a downgrade to CUDA 9.0 is requested

Yea, it'd be great if the ethminer could remain compatible with older Nvidia drivers (especially since new driver is still not quite ready to be used in production environment).

[StefanOberhumer]

Can't run that command as there is no ./ethminer folder in smOS (at least I can't find it).

As I remember the miners are at /root/miner/ethminer.... or at /root/miner_org/ethminer....

You also can adapt you commandline (in the SMOS interface) to -U --list-devices