Suggestion for your script: “killall -q optiminer-zcash”
You don’t need to do all the “ps” stuff.
1 Like
the scripting is sooo incomprehensible to me.
I wish that someone just post the actual code (uncommented and formatted so each line actually represents a line of code). That would be greatly appreciated 
PS: this mb apparently has no real ability to post the code (vs ethereum forum), hence the confusion
First, what are you looking for? The watchdog-cmd.sh? For linux?
Secondly, what are you trying to accomplish? Miner restart? System reboot? Hard reset? A love poem of the day?
to unstuck the Optiminer, however, since on my rigs optiminer 1.6 always results in unresponsive system (not seg fault), i am not sure that it is even possible, but any help is appreciated on “Miner restart” on Linux.
PS Love poem? Maybe if we were in “Westworld”.
If you are using the mine.sh script in the optiminer install directory, it will automatically restart the miner if it quits. Have you looked at it? I am not sure if that’s what you are looking for or not.
I am not using any script, was just commenting on a script that someone had posted, which was absolutely not clear to me since i cannot distinguish where the actual command begins and comment ends or vice versa.
1.6.1 and 1.6.2 freeze my ubuntu 16.04 system.
Does it mean that mine.sh is not active and I somehow need to make it run?
I was looking for something that can overcome this freeze, if possible.
Thanks
First, the script, above, that you are referring to finds the process id of the optiminer process and kills it, then reboots the system. The reason you can’t read it is that you have to click on the “V” button to expand the view, at which point the posting becomes completely legible. HTH.
2 Likes
Just go to the main post.
So I’ve one rig of 5 Ellesmere cards running optiminer 1.6.2 on Ubuntu 16.04 server without any X cruft. AMD 16.40 drivers and v3 of the SDK.
All of the cards do ~300 sol/s except for GPU0. It does 150 sol/s when a monitor is plugged into it, and 62 sol/s when I boot headless.
I have integrated graphics turned off.
Any ideas how to fix it?
------- Device list -------
foo@bar:~/optiminer-zcash$ ./optiminer-zcash --list-devices
[2017-02-17 14:30:50.648] [info] Optiminer/Zcash 1.6.2 (C) Optiminer 2017
Platform (-c) 0, Device (-d) 0:
Platform Vendor: AMD
Device Name: Ellesmere
Driver Version: 2117.10 (VM)
Address bits: 64
Available memory: 1691627520
GCN Support: Gen 4
Has device specific faster kernel: Yes
Platform (-c) 0, Device (-d) 1:
Platform Vendor: AMD
Device Name: Ellesmere
Driver Version: 2117.10 (VM)
Address bits: 64
Available memory: 5981958144
GCN Support: Gen 4
Has device specific faster kernel: Yes
Platform (-c) 0, Device (-d) 2:
Platform Vendor: AMD
Device Name: Ellesmere
Driver Version: 2117.10 (VM)
Address bits: 64
Available memory: 5981958144
GCN Support: Gen 4
Has device specific faster kernel: Yes
Platform (-c) 0, Device (-d) 3:
Platform Vendor: AMD
Device Name: Ellesmere
Driver Version: 2117.10 (VM)
Address bits: 64
Available memory: 5981958144
GCN Support: Gen 4
Has device specific faster kernel: Yes
Platform (-c) 0, Device (-d) 4:
Platform Vendor: AMD
Device Name: Ellesmere
Driver Version: 2117.10 (VM)
Address bits: 64
Available memory: 5981958144
GCN Support: Gen 4
Has device specific faster kernel: Yes
Another oddity with an Ellesmere card on another rig. Shows up with --list-devices, and mining will start. However, it never completes any work and Optiminer hangs when I do a ctrl-c.
The card works fine for a vga display, but it doesn’t seem to be able to do any work. It’s the same model as I showed in my previous message.
Ideas?
I am curious whether AMD-APP-SDK and/or ADL_SDK is/are needed for Optiminer?
I have them installed on some rigs and maybe not others.
Besides, what would be a procedure to remove 2.9 and install 3.0?
Just remove the directory?
What if I have both 2.9 and 3.0 installed?
No they are not. All the OpenCL stuff is included in the basic driver install. The only thing that the SDK has is the header files, which are needed to build in SDK sample code. Since you aren’t building any GPU kernel code, there’s no need to install the SDK, and it usually buggers up your library path and such.
Have you tried swapping risers, if they are on risers. Sounds a lot like a flakey riser or MB slot issue to me.
I’ll give that a shot. It was working fine previously, but I guess a riser could flake out. Thanks!
You start a optiminer for each card with the GPU index given on the command line:
optiminer-zcash -d 0 -s nexus:9999 -u node01 -p x --watchdog-timeout 6 --watchdog-cmd ./watchdog-cmd
optiminer-zcash -d 1 -s nexus:9999 -u node01 -p x --watchdog-timeout 6 --watchdog-cmd ./watchdog-cmd
optiminer-zcash -d 2 -s nexus:9999 -u node01 -p x --watchdog-timeout 6 --watchdog-cmd ./watchdog-cmd
optiminer-zcash -d 3 -s nexus:9999 -u node01 -p x --watchdog-timeout 6 --watchdog-cmd ./watchdog-cmd
optiminer-zcash -d 4 -s nexus:9999 -u node01 -p x --watchdog-timeout 6 --watchdog-cmd ./watchdog-cmd
optiminer-zcash -d 5 -s nexus:9999 -u node01 -p x --watchdog-timeout 6 --watchdog-cmd ./watchdog-cmd
I replaced risers, but no joy. It’s just odd that a card will work for video, but hangs with optiminer. My other cards of same type have no issues.
Your --list-device output above shows that card 0 has much fewer available memory. So, either something is using up that card’s memory or something is broken with the card. Have you tried switching pci positions?
I did. Same problem.
Try setting intensity 1.
You say the same card works fine with v1.5?
So I did a bit more diagnoses:
The card in question works fine with 1.6.0, although it does run at 1/2 the speed of others similar to another rig. That’s likely because it’s the one being used as the display.
Evidently there’s a change between 1.6.0 and 1.6.2 that hangs that card.