Update_Core, again

[Roland140]

I'm stuck with this one. I have been folding for a while using Win7/64/Pro and just upgraded(?) to Win10/64/Pro. I am not a fan of Win10 but am stuck with it because support for a vital piece of software ends soon for Win7. Anyhow, the original install of F@H under Win10 worked but there was a problem with the video driver not seeing one card or all options on the other two. I reinstalled the latest and greatest directly from EVGA and haven't been able to get F@H working again.

I've manually added rules to Defender for F@H as well as accepted the rule updates when prompted. I even added a rule for incoming port 49995 trying to get the core update. Nothing has worked. There are tufts of hair on my desk from me pulling it out, and I'm bald! Any help would be appreciated.

Machine - i7-6700K in an Asus ROG Maximus VIII
Memory - 32GB
Video cards - RTX2070, GTX1060, GTX970
Video driver - 419.35
F@H - 7.5.1
Internet - Comtrash

Code: Select all

*********************** Log Started 2019-03-25T15:46:18Z ***********************
15:46:18:************************* Folding@home Client *************************
15:46:18:        Website: https://foldingathome.org/
15:46:18:      Copyright: (c) 2009-2018 foldingathome.org
15:46:18:         Author: Joseph Coffland <joseph@cauldrondevelopment.com>
15:46:18:           Args: 
15:46:18:         Config: C:\Users\Roland\AppData\Roaming\FAHClient\config.xml
15:46:18:******************************** Build ********************************
15:46:18:        Version: 7.5.1
15:46:18:           Date: May 11 2018
15:46:18:           Time: 13:06:32
15:46:18:     Repository: Git
15:46:18:       Revision: 4705bf53c635f88b8fe85af7675557e15d491ff0
15:46:18:         Branch: master
15:46:18:       Compiler: Visual C++ 2008
15:46:18:        Options: /TP /nologo /EHa /wd4297 /wd4103 /Ox /MT
15:46:18:       Platform: win32 10
15:46:18:           Bits: 32
15:46:18:           Mode: Release
15:46:18:******************************* System ********************************
15:46:18:            CPU: Intel(R) Core(TM) i7-6700 CPU @ 3.40GHz
15:46:18:         CPU ID: GenuineIntel Family 6 Model 94 Stepping 3
15:46:18:           CPUs: 8
15:46:18:         Memory: 31.94GiB
15:46:18:    Free Memory: 27.37GiB
15:46:18:        Threads: WINDOWS_THREADS
15:46:18:     OS Version: 6.2
15:46:18:    Has Battery: false
15:46:18:     On Battery: false
15:46:18:     UTC Offset: -4
15:46:18:            PID: 10668
15:46:18:            CWD: C:\Users\Roland\AppData\Roaming\FAHClient
15:46:18:             OS: Windows 10 Enterprise
15:46:18:        OS Arch: AMD64
15:46:18:           GPUs: 3
15:46:18:          GPU 0: Bus:2 Slot:0 Func:0 NVIDIA:7 GP106 [GeForce GTX 1060 3GB] 3935
15:46:18:          GPU 1: Bus:6 Slot:0 Func:0 NVIDIA:5 GM204 [GeForce GTX 970]
15:46:18:          GPU 2: Bus:1 Slot:0 Func:0 NVIDIA:7 TU106 [GeForce RTX 2070 Rev. A] M
15:46:18:                 7465
15:46:18:  CUDA Device 0: Platform:0 Device:0 Bus:1 Slot:0 Compute:7.5 Driver:10.1
15:46:18:  CUDA Device 1: Platform:0 Device:1 Bus:2 Slot:0 Compute:6.1 Driver:10.1
15:46:18:  CUDA Device 2: Platform:0 Device:2 Bus:6 Slot:0 Compute:5.2 Driver:10.1
15:46:18:OpenCL Device 0: Platform:0 Device:0 Bus:1 Slot:0 Compute:1.2 Driver:419.35
15:46:18:OpenCL Device 1: Platform:0 Device:1 Bus:2 Slot:0 Compute:1.2 Driver:419.35
15:46:18:OpenCL Device 2: Platform:0 Device:2 Bus:6 Slot:0 Compute:1.2 Driver:419.35
15:46:18:  Win32 Service: false
15:46:18:***********************************************************************
15:46:18:<config>
15:46:18:  <!-- HTTP Server -->
15:46:18:  <allow v='192.168.132.0/0'/>
15:46:18:
15:46:18:  <!-- Network -->
15:46:18:  <proxy v=':8080'/>
15:46:18:
15:46:18:  <!-- Remote Command Server -->
15:46:18:  <command-allow-no-pass v='192.168.132.0/0'/>
15:46:18:
15:46:18:  <!-- Slot Control -->
15:46:18:  <power v='FULL'/>
15:46:18:
15:46:18:  <!-- User Information -->
15:46:18:  <passkey v='********************************'/>
15:46:18:  <team v='223274'/>
15:46:18:  <user v='awsum140'/>
15:46:18:
15:46:18:  <!-- Folding Slots -->
15:46:18:  <slot id='4' type='GPU'>
15:46:18:    <client-type v='advanced'/>
15:46:18:    <cuda-index v='0'/>
15:46:18:    <gpu-index v='2'/>
15:46:18:    <opencl-index v='0'/>
15:46:18:  </slot>
15:46:18:  <slot id='0' type='GPU'>
15:46:18:    <client-type v='advanced'/>
15:46:18:    <cuda-index v='1'/>
15:46:18:    <gpu-index v='0'/>
15:46:18:    <opencl-index v='1'/>
15:46:18:  </slot>
15:46:18:  <slot id='1' type='GPU'>
15:46:18:    <client-type v='advanced'/>
15:46:18:    <cuda-index v='2'/>
15:46:18:    <gpu-index v='1'/>
15:46:18:    <opencl-index v='2'/>
15:46:18:  </slot>
15:46:18:</config>
15:46:18:Trying to access database...
15:46:19:Successfully acquired database lock
15:46:19:Enabled folding slot 04: READY gpu:2:TU106 [GeForce RTX 2070 Rev. A] M 7465
15:46:19:Enabled folding slot 00: READY gpu:0:GP106 [GeForce GTX 1060 3GB] 3935
15:46:19:Enabled folding slot 01: READY gpu:1:GM204 [GeForce GTX 970]
15:46:19:WU03:FS04:Downloading core from http://cores.foldingathome.org/Win32/AMD64/NVIDIA/Fermi/Core_21.fah
15:46:19:WU03:FS04:Connecting to cores.foldingathome.org:80
15:46:20:ERROR:WU03:FS04:Exception: Failed reading core package header.
15:47:19:WU03:FS04:Downloading core from http://cores.foldingathome.org/Win32/AMD64/NVIDIA/Fermi/Core_21.fah
15:47:19:WU03:FS04:Connecting to cores.foldingathome.org:80
15:47:19:ERROR:WU03:FS04:Exception: Failed reading core package header.
15:48:56:WU03:FS04:Downloading core from http://cores.foldingathome.org/Win32/AMD64/NVIDIA/Fermi/Core_21.fah
15:48:56:WU03:FS04:Connecting to cores.foldingathome.org:80
15:48:56:ERROR:WU03:FS04:Exception: Failed reading core package header.
15:51:33:WU03:FS04:Downloading core from http://cores.foldingathome.org/Win32/AMD64/NVIDIA/Fermi/Core_21.fah
15:51:33:WU03:FS04:Connecting to cores.foldingathome.org:80
15:51:33:ERROR:WU03:FS04:Exception: Failed reading core package header.
15:55:48:WU03:FS04:Downloading core from http://cores.foldingathome.org/Win32/AMD64/NVIDIA/Fermi/Core_21.fah
15:55:48:WU03:FS04:Connecting to cores.foldingathome.org:80
15:55:48:ERROR:WU03:FS04:Exception: Failed reading core package header.
16:02:39:WU03:FS04:Downloading core from http://cores.foldingathome.org/Win32/AMD64/NVIDIA/Fermi/Core_21.fah
16:02:39:WU03:FS04:Connecting to cores.foldingathome.org:80
16:02:39:ERROR:WU03:FS04:Exception: Failed reading core package header.

[bruce]

You may (or may not) have two problems. The primary one you're asking about doesn't seem to be a problem for me with Windows7 or 10, so I can't really answer your primary question at this time. (Maybe that will no longer be a problem once you take care of the problem I do understand.)

I notice that all 6 of your slots are attempting to download the same core at the same time. What I can tell you is that FAH has a problem when more than one slot tries to download the same FAHCore at the same time. Windows will attempt to write all 6 copies in the exact same place and at least 5 will fail because that file can only be open once for writing. The work-around is for you to manually pause all GPU slots and then enable one of them.

(I recommend you set pause-on-start = true and restart FAH and then manually start one slot. There are other ways that this might be accomplished.) Once FAHCore_21 has been downloaded, it will be available for the other slots to use and the other 5 copies will not be needed.

At some later time, you may receive a WU which requires a new FAHCore, but the likelihood is that it will only happen for single slots, one-at-a-time.

Let us know if this works, and if not, we can look into your other problem.

[bruce]

You only have 3 GPUs, not 6, so manually constructing 3 extra slot is incorrect.

A single GPU cannot process two WUs, even if you have extra slots to manage those extra WUs.
The GPU can process an OpenCl assignment OR a CUDA assignment, not both, and the same slot would be used for either.

At the present time, there are no CUDA-based WUs, so you will be assigned 3 OpenCL WUs using either FAHCore_21 or maybe FAHCore_22

15:46:18: GPUs: 3
15:46:18: GPU 0: Bus:2 Slot:0 Func:0 NVIDIA:7 GP106 [GeForce GTX 1060 3GB] 3935
15:46:18: GPU 1: Bus:6 Slot:0 Func:0 NVIDIA:5 GM204 [GeForce GTX 970]
15:46:18: GPU 2: Bus:1 Slot:0 Func:0 NVIDIA:7 TU106 [GeForce RTX 2070 Rev. A] M

[Roland140]

I went one better. I removed two slots and re-booted the machine. Same error.

Is there a way to manually download the cores?

[Joe_H]

As long as the message is "Failed reading core package header." something is blocking your download of the core. The F@h client uses an HTTP connection over either port 8080 or port 80. It has to have the same access as a browser, it looks like the rules you set for Defender are not right or there is something else that also needs an exception created for the FAHClient process.

There is no way currently to download the cores manually. The compressed file containing the core gets downloaded, unpacked and properly located in the directory by the FAHClient process.

[Roland140]

I just went even further. I installed Norton and added F@H to the list there. Still nothing. Did I mention I hate Windoohs 10?

I ran WireShark and it looks like cores is trying to use port 49995, not 80 or 8080. I understand that it is being blocked, somehow, by something but this is the first time I've ever had a problem getting F@H to run after running it for years on various builds with various OS. Very frustrating to have a machine that produced 1.5-2 million points per day sitting here like a boat anchor.

[bruce]

Can you open http://cores.foldingathome.org/Win32/AM ... ore_21.fah in your browser? If not, what's blocking it?

If so, it was just a test and the downloaded file is unneeded.

[Roland140]

Popped right on the screen as downloaded, Bruce.

I'm starting to wonder if it's a permissions problem on the AppData directory. You wouldn't think so since F@H created the base structure, but seems like it's having a problem going further.

[bruce]

it is guaranteed to have a problem unless you use the shortcut. What CWD directory is being used?

[bruce]

Re-post your log showing the segment before after starting one slot and the part as you enable that slot.

I'm assuming you enabled FAHClient.exe to have the same internet connection permissions as a browser.

20:30:13:WU01:FS02:Connecting to 65.254.110.245:8080
20:30:14:WU01:FS02:Assigned to work server 155.247.166.220
20:30:14:WU01:FS02:Requesting new work unit for slot 02: READY gpu:1:GM206 [GeForce GTX 960] 2308 from 155.247.166.220
20:30:14:WU01:FS02:Connecting to 155.247.166.220:8080
20:30:15:WU01:FS02:Downloading 629.02KiB
20:30:16:WU01:FS02:Download complete
20:30:16:WU01:FS02:Received Unit: id:01 state:DOWNLOAD error:NO_ERROR project:14166 run:13 clone:90 gen:0 core:0x21 unit:0x000000000002894c5c38c06b4ecadd5f
20:30:16:WU01:FS02:Downloading core from http://cores.foldingathome.org/Win32/AM ... ore_21.fah
20:30:16:WU01:FS02:Connecting to cores.foldingathome.org:80
20:30:16:WU01:FS02:FahCore 21: Downloading 3.47MiB
20:30:17:WU01:FS02:FahCore 21: Download complete
20:30:18:WU01:FS02:Valid core signature
20:30:18:WU01:FS02:Unpacked 11.80MiB to cores/cores.foldingathome.org/Win32/AMD64/NVIDIA/Fermi/Core_21.fah/FahCore_21.exe
20:30:18:WU01:FS02:Starting
20:30:18:WU01:FS02:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:\Users\bruce\AppData\Roaming\FAHClient\cores/cores.foldingathome.org/Win32/AMD64/NVIDIA/Fermi/Core_21.fah/FahCore_21.exe -dir 01 -suffix 01 -version 705 -lifeline 5256 -checkpoint 15 -gpu-vendor nvidia -opencl-platform 0 -opencl-device 0 -cuda-device 0 -gpu 0
20:30:18:WU01:FS02:Started FahCore on PID 7240
20:30:19:WU01:FS02:Core PID:5908
20:30:19:WU01:FS02:FahCore 0x21 started

[Roland140]

Thanks for trying to help, Bruce. Sorry for the delay, dinner time.

What is the "CWD" directory? You've got me with that term. F@H has the same access to the network/internet as a browser. I also added fahcorewrapper.exe to the list, with the same permissions, but no dice.

Here's the log. I deleted the GPUs and created a new one to get a fresh log file.

Code: Select all

21:59:44:Adding folding slot 00: READY gpu:2:TU106 [GeForce RTX 2070 Rev. A] M 7465
21:59:44:Saving configuration to config.xml
21:59:44:<config>
21:59:44:  <!-- Network -->
21:59:44:  <proxy v=':8080'/>
21:59:44:
21:59:44:  <!-- Slot Control -->
21:59:44:  <power v='full'/>
21:59:44:
21:59:44:  <!-- Folding Slots -->
21:59:44:  <slot id='0' type='GPU'>
21:59:44:    <cuda-index v='0'/>
21:59:44:    <gpu-index v='2'/>
21:59:44:    <opencl-index v='0'/>
21:59:44:  </slot>
21:59:44:</config>
21:59:44:WU00:FS00:Connecting to 65.254.110.245:8080
21:59:44:WU00:FS00:Assigned to work server 155.247.166.220
21:59:44:WU00:FS00:Requesting new work unit for slot 00: READY gpu:2:TU106 [GeForce RTX 2070 Rev. A] M 7465 from 155.247.166.220
21:59:44:WU00:FS00:Connecting to 155.247.166.220:8080
21:59:47:WU00:FS00:Downloading 835.91KiB
21:59:47:WU00:FS00:Download complete
21:59:47:WU00:FS00:Received Unit: id:00 state:DOWNLOAD error:NO_ERROR project:14166 run:12 clone:36 gen:9 core:0x21 unit:0x0000000a0002894c5c38c06238990bbe
21:59:58:Saving configuration to config.xml
21:59:58:<config>
21:59:58:  <!-- Network -->
21:59:58:  <proxy v=':8080'/>
21:59:58:
21:59:58:  <!-- Slot Control -->
21:59:58:  <power v='full'/>
21:59:58:
21:59:58:  <!-- Folding Slots -->
21:59:58:  <slot id='0' type='GPU'>
21:59:58:    <cuda-index v='0'/>
21:59:58:    <gpu-index v='2'/>
21:59:58:    <opencl-index v='0'/>
21:59:58:  </slot>
21:59:58:</config>
22:19:28:WU00:FS00:Downloading core from http://cores.foldingathome.org/Win32/AMD64/NVIDIA/Fermi/Core_21.fah
22:19:28:WU00:FS00:Connecting to cores.foldingathome.org:80
22:19:28:ERROR:WU00:FS00:Exception: Failed reading core package header.

It's interesting to note the time lag from downloading the work unit to the time it tries to get the core, over 20 minutes. Wonder why that is?

[bruce]

The same code is used (more or less) for both Linux and Windows. The CWD (Change Working Directory) ins the Linux equivalent of Windows CD command (Change Directory) and in this case, it reports the current working direotry where the program which is beomg executed will attempt to create files and/or directories. (Unless more detail is given.) FAH will download an assigned WU there and temporary files will be created there (potentially modified in some way by a file structure than is created within that location.

[Roland140]

The work directory is C:\users\roland\appdata\roaming\fahclient\work. Under that is "00" for the work unit that was downloaded. In the 00 directory there is only the wudata_01 file, nothing else, while on the working machine there are additional files and an 01 directory, of course, since it's working. Same as on my Win10/64/Pro laptop where F@H works fine.

[Roland140]

I was getting desperate. I went looking around and found a copy of the original directory structure from Win10. Being the cautious type I cloned the drive before I "operated" on it again. I copied the cores directory from that, the cloned copy, install into the new install and everything is running fine. The root problem still exists though, which worries me.

[bruce]

Copying the cores from one machine to another (as log as you preserve the directory structure) should work. Copying ANY other files is a bad idea.

But as you say, it doesn't solve the original problem.

If you can download the FAHCore with you browser, you have not successfully provided the proper firewall exception for the program FAHClient. FAHClient is not permitted to act as a browser. Whether you're potected by a Windows firewall or a Norton Firewall, they once were set up to warn you when they blocked something .... but most people disable those messages because they get too many of them.