F&H stuck in download

[acnash]

Hi all,

Sorry if this has been mentioned elsewhere, still quite new to navigating the forum. My Linux distribution of F&H has sat in "Download" for the past six hours, with no progress. ETA unknown, Credit unknown, PRCG 0, work server 155.247.166.219.

I've stopped, restarted, paused, played etc., with no change and no mention of unavailable workunits. No errors in the log. My GPU is running fine, it's just the CPU jobs.

Thanks
Anthony

[Joe_H]

There is a known bug in the folding client where a download or upload fails or stalls and the client does not detect the failure. The slot will stay in that state until the FAHClient process is restarted. The easiest way to restart FAHClient is rebooting the system, you can also restart it by issuing a restart command from the command line after pausing your other folding slots.

The network bug is more of a problem with the release version of the client, 7.4.4. The public beta 7.4.16 has some improvements in the network detection code and will restart the stalled download or upload most of the time, but it can take 15-30 minutes.

[acnash]

Thanks for the reply, apologies but could you 'walk' me through how to issue a restart command from the command line - is this something F&H specific or do you mean a restart of my linux box? I'm sharing my box with some of my DFT calculations at the moment so a complete system restart would have to wait a few days.

Thanks for your help.

[Joe_H]

I usually do this when needed on OS X, so I had to look up the correct location for Linux. From the command line use the following command:

Code: Select all

sudo /etc/init.d/FAHClient restart

you will be prompted for an admin password.

[SteveWillis]

Since you are Linux I'll mention that I have a script (reboot.sh) that monitors for several conditions including a stuck download and performs appropriate actions, in this case restarts the client. You can download it if you are interested. It is, however a work in progress.
https://drive.google.com/drive/folders/ ... sp=sharing

[acnash]

Hi Joe_H,
I restarted the client, nothing happened to the slots on the FAHControl so closed that down and loaded it back up. Oddly the FAHControl on restarting has gone into an "Updating Inactive" state. Its process is marked S and nload is showing no internet traffic (so not updating I assume). This is version 7.4.4. I guess an upgrade to 7.4.16 might help? Sorry for all the basic question, I'm a comp chem researcher so my Linux knowledge is just enough to get me through without being fluent. What are the cleanest means of upgrading the Controler?

Hi SteveWillis, thanks for that. I will be abroad from this box for almost a month so having something like your script would be very useful.

Thanks both.

[bruce]

-FAH'Client was designed specifically to be run as a daemon started at boot-time by that script, not started by any other method.

There's a very high probability that you have created a permissions problem by starting FAHClient as your userid (with or without sudo) by some method OTHER THAN using that script. FAHClient uses different Linux user permissions than your user does. The directory containing FAH's working files is not supposed to belong to the permission-set of your user.

[Bedrich_Hajek]

This happened to me about 2 weeks ago. To solve it, I had to completely uninstall fah 7.4.4, and then reinstall it. It has been running fine for a few days. I managed to complete 60 WUs successfully so far, but I am not getting any bonus points.

[bruce]

Hopefully it's a simple as entering your name and passkey into FAHControl -- preferably the same two that you used previously to earn bonus points.

[acnash]

Hi Bruce,
Quite right. I didn't know it was a daemon and I bodged up the permissions. A complete restart of the system resolved the problem. F@H is currently working.

I'll look into the script provided, although they are probably a little beyond where I am being just a mint Linux user.

[SteveWillis]

If you are going to be away for awhile you might also do what I do. I go into my motherboards bios and there is a setting to cause the pc to automatically boot when power is restored after a power outage. It can be pretty hard to find but I imagine most motherboards have it. It would be a shame if your folding went down due to a one second power outage. Also, if you will have a laptop or tablet you can use nomachine (nomachine.com) to connect to and check your pc while you are gone. It's free and easy to set up and works great.

[Bedrich_Hajek]

Hopefully it's a simple as entering your name and passkey into FAHControl -- preferably the same two that you used previously to earn bonus points.

Yes, this worked! Thanks.

[acnash]

Bonus points!? I'm yet to read about bonus points.

As things stand I have a functioning version 7.4.4 controller. How do I upgrade to 7.4.16 to resolve the issue of a perpetual stalled download?

Then I'll look into the automatic boot when power is restored, although that would surely just bring me to a username prompt. How would I get around that?

[Joe_H]

Announcement of the public beta version is here - viewtopic.php?f=104&t=29541. Download from the Beta page - https://folding.stanford.edu/beta/.

There are a few known issues, mention has been added to the announcement post. Basically, the default theme does not work right on Windows and there is a memory leak in FAHControl. The workaround for that is to just open FAHControl when needed to pause, start or otherwise monitor and configure the folding client.

As for a restart, the default settings for the folding client do not require that you be logged in for the client to run and process WU's. With FAHClient running in the background as a daemon, it will start processing any WU's again at the last checkpoint.

[SteveWillis]

I really don't know much about windows but in Linux folding starts at boot, not when you log in. So folding starts at the log in prompt even before you log in.