Trouble installing on Ubuntu & some other issues

[Azmodes]

I want to get F@H running on all my crunchers normally doing BOINC, but been having some problems with that since most of them are running Ubuntu. I've been trying to follow the instructions here, but when installing the client package on two machines running Ubuntu 18.04 and 19.10 respectively it just says

Errors were encountered while processing:
fahclient

...and nothing more, which is not exactly enlightening. There's a /var/lib/fahclient/ being created, but it's empty.

On another computer with 18.04, I can get it to install, but I simply am unable to start fahcontrol. It installs and a shortcut shows up for executing, but it just doesn't open anything and there's nothing in the running processes. I can see that the client is working in the background, but I have no way to control anything (and it's only doing CPU work, no GPUs).

Unrelated, on my Windows machines, I can run everything fine and the client/control thing works, but I can't connect to the web control page. It keeps refreshing and refreshing and can't connect. Minor issue, I guess.

Also been having on and off problems with getting a consistent stream of work from the server, but I assume that's because of a lack of work in light of a presumed influx of new users due to covid-19 work units.

Thanks in advance.

[totlium]

See if this helps:

Code: Select all

sudo update-rc.d FAHClient defaults

[foldy]

The linux postinstall is broken on Ubuntu. So after install I clean up using

Code: Select all

rm /var/lib/dpkg/info/fahclient.postinst
dpkg --configure -a

This is how I start the FahClient without FAHcontrol

Code: Select all

cd /etc/init.d/
FAHClient

To edit the config manually

Code: Select all

cd /etc/init.d/
nano config.xml

[Azmodes]

See if this helps:

Code: Select all

sudo update-rc.d FAHClient defaults

That seems to have worked. However, it looks like FAHcontrol needs certain Python packages that my version of Ubuntu doesn't support anymore?

This is how I start the FahClient without FAHcontrol

Code: Select all

cd /etc/init.d/
FAHClient

To edit the config manually

Code: Select all

cd /etc/init.d/
nano config.xml

This also worked, but how can I get it to crunch on the GPU(s)? What do I add or modify in the config file? Do I just add new instances for "GPU" to the <slot>, with new IDs?

[totlium]

Did you install python-gnome2?

[Azmodes]

I tried, but

Code: Select all

Package python-gnome2 is not available, but is referred to by another package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source

E: Package 'python-gnome2' has no installation candidate

Also, I attempted to modify the config file with some info I found elsewhere, but my two graphic cards aren't even being recognized:

Code: Select all

5:34:53:******************************* System ********************************
15:34:53:        CPU: Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz
15:34:53:     CPU ID: GenuineIntel Family 6 Model 58 Stepping 9
15:34:53:       CPUs: 8
15:34:53:     Memory: 23.19GiB
15:34:53:Free Memory: 15.87GiB
15:34:53:    Threads: POSIX_THREADS
15:34:53: OS Version: 5.3
15:34:53:Has Battery: false
15:34:53: On Battery: false
15:34:53: UTC Offset: 1
15:34:53:        PID: 28424
15:34:53:        CWD: /etc/init.d
15:34:53:         OS: Linux 5.3.0-29-generic x86_64
15:34:53:    OS Arch: AMD64
15:34:53:       GPUs: 0
15:34:53:       CUDA: 7.5
15:34:53:CUDA Driver: 10020
15:34:53:***********************************************************************
15:34:53:<config>
15:34:53:  <!-- Slot Control -->
15:34:53:  <power v='full'/>
15:34:53:
15:34:53:  <!-- Folding Slots -->
15:34:53:  <slot id='0' type='GPU'/>
15:34:53:  <slot id='1' type='GPU'/>
15:34:53:</config>
15:34:53:Connecting to assign-GPU.stanford.edu:32563
15:37:03:WARNING:Attempting to update GPUs.txt from assign-GPU.stanford.edu:80: Failed to connect to assign-GPU.stanford.edu:32563: Connection timed out
15:37:03:Connecting to assign-GPU2.stanford.edu:32563
15:37:03:WARNING:Attempting to update GPUs.txt from assign-GPU2.stanford.edu:80: Failed to connect to assign-GPU2.stanford.edu:32563: Connection refused
15:37:03:Trying to access database...
15:37:03:Successfully acquired database lock
15:37:03:ERROR:Exception: No available GPUs
15:37:03:ERROR:Exception: No available GPUs
15:37:03:ERROR:No valid folding configuration
15:37:03:ERROR:Exception: Option 'gpu-index' has no default and is not set.
15:37:04:ERROR:WU00:FS00:Exception: Option 'gpu-index' has no default and is not set.
15:37:04:ERROR:WU01:FS01:Exception: Option 'gpu-index' has no default and is not set.

They work fine for various BOINC projects. It's two 2070 Supers, driver 440.44, OCL package is installed.

[totlium]

I am new to this myself - I saw somewhere there's a GPU whitelist. See if you cards are included there.

[Azmodes]

viewtopic.php?f=83&t=24249 This one? Looks a bit dated.

But I'm seeing posts by other people folding with their 2070 Supers, so I guess it must be whitelisted. Would be strange if it weren't, anyway.

[foldy]

It cannot download the GPU whitelist GPUs.txt
Maybe the server is temporarily not available or you have some firewall blocking assign-GPU.stanford.edu?

[Azmodes]

I don't think it's a firewall issue and considering the current server issues(?) maybe it's not on my end.

EDIT: Well, the computer in question can access assign-GPU.stanford.edu just fine.

[Azmodes]

oh by the way, I tried the instructions in this thread (should've checked before making this one maybe...) on another computer and it recognizes the GPUs (1080 Ti and 1070 Ti), but I get the same error messages as totlium. Failed to get assignment etc.

EDIT: ah, just received some work, although only for one card.

[Joe_H]

I don't think it's a firewall issue and considering the current server issues(?) maybe it's not on my end.

EDIT: Well, the computer in question can access assign-GPU.stanford.edu just fine.

There are two different issues that you can run into with a firewall or other anti-malware app. First, the site can just plain be blocked. Second, access over HTTP(S) can be restricted to just "known browsers" such as Firefox, Chrome, Opera and so on.

In the second case an exception will need to be added for the FAHClient process to use HTTP(S) over ports 8080 and/or 80 so FAHCleint can connect to the servers.

[Azmodes]

I briefly disabled ufw and the issue persists; no GPUs detected, can't connect to assign-GPU.stanford.edu:32735.

[Vocylis]

I am having some issues installing fahcontrol on Ubuntu 19.10 as well and I think I know why. It appears the package for python-gnome2 does not exist in Eoan 19.10, but it does exist for Bionic 18.04.

If you click the links to the Ubuntu package website (can't link here because new user, sorry!) and then click "Eoan" > "Python", you can see that there the package in question is listed, but go to "Bionic" > "Python" instead and it's missing. Unfortunately, I don't think Folding@home's team has noticed this yet. I'm not even sure how I can let them know about this... I don't see an appropriate place in the forums where they would browse nor do I see a contact method.

[Azmodes]

I don't actually mind the inability to run fahcontrol that much. I can handle things well enough via the config file and the terminal. But I still cannot get my 19.xx systems to recognize their GPUs.