Page 1 of 1
Fails to restart after disconnection from internet
Posted: Sun Sep 20, 2020 11:43 pm
by jarruza
F@H was running, I disconnected the internet, F@H was working on a WU. Now it is not running. I reconnected the internet expecting the process to automatically start up. However, F@H is still in inactive state and did not restart/upload the WU it completed. I still see the process in Taskmgr but at 0% CPU.
As anyone seen this behavior before, i.e., disconnecting the internet while processing a WU causes F@H to eventually stop working even after reconnecting? I would expect that it would periodically check for internet connectivity then upload the results and pull down the next WU.
Thanks
Re: Fails to restart after disconnection from internet
Posted: Mon Sep 21, 2020 1:25 am
by JimboPalmer
Welcome to Folding@Home!
When a Work Unit finishes, the Client starts looking for another WU. At first it tries fairly often, but over time it checks less and less often.
You can speed this up by stopping and restarting the client or just rebooting the PC.
Re: Fails to restart after disconnection from internet
Posted: Mon Sep 21, 2020 1:26 pm
by jarruza
If I reboot the PC, will I loose the work or is it saved? If it by default it's lost, is there a way to preserve the work, e.g., pause the app, reboot then start (or unpause)?
Thanks
Re: Fails to restart after disconnection from internet
Posted: Mon Sep 21, 2020 1:53 pm
by Neil-B
If you pause the slots, wait a minute or so then pause you should reset the retry timer.
Re: Fails to restart after disconnection from internet
Posted: Mon Sep 21, 2020 2:21 pm
by JimboPalmer
jarruza wrote:If I reboot the PC, will I loose the work or is it saved?
I seem to have misunderstood your situation.
I thought you were completing Work Units while not on line and then when you become online, you do not upload that work.
Now you ask about work in the middle of being processed, so my image of what is happening is flawed.
Can you explain your set up?
Re: Fails to restart after disconnection from internet
Posted: Mon Sep 21, 2020 2:34 pm
by Knish
if it is a CPU WU, you won't lose it through a reboot. If it is a GPU WU, there is a chance that if you do not have an internet connection up and running when FAHClient starts that if the client tries to update the GPUs.txt file it will drop the WU data. This only happens about once a month though. Example: I have a PC with a wifi dongle that I have to manually connect to the internet. I left this pc folding for more than a month straight, then needed to reboot. I paused after letting all WUs finish. After reboot, since my fahclient started up before I could connect my wifi, it tried to update GPUs.txt and failed, thus couldn't verify the GPU slot, so dropped the GPU WU. the cpu WU was fine tho.
Re: Fails to restart after disconnection from internet
Posted: Mon Sep 21, 2020 4:43 pm
by jarruza
I seem to have misunderstood your situation.
I thought you were completing Work Units while not on line and then when you become online, you do not upload that work.
No misunderstanding, the WU was completed while the device was offline. Once back online F@H did not upload.
However, based on the comments in this thread, I feel I may be able to overcome this shortcoming programmatically by checking for internet connectivity and if it's not connected, say for 15/30 minutes, send the pause command to F@H then wait for internet connectivity and once established, unpause F@H.
Thanks
Re: Fails to restart after disconnection from internet
Posted: Tue Sep 22, 2020 3:50 am
by PantherX
jarruza wrote:...the WU was completed while the device was offline. Once back online F@H did not upload...
If you look at the log file, you may notice that once the WU was finished, it tried to upload and failed which is expected. However, the retry for the next upload is using an exponential back-off function which caps it to 1 hour maximum between retry. The fastest way to recover from that is to either pause/unpause your slot or restart the client.