Project 9201 WUs not finishing

[ThePegasus1979]

Dear FAH team,

since 2 days i get project 9201 WUs for my GPU only (Sapphire R390 Nitro+, up to date AMD Crimson drivers, Windows 7.1 64 Bit with all update). The WUs over and over again get stuck at 99.99%, no finishing, stating a remaining calculation time of 10 seconds. After waiting 2 hours without a change in progress (from 0% to 99.99% it took 3,5 hours) i stop the client (which takes ages as 9201 seems to be really nasty bugged), delete the corresponding "work" subfolder, restart the client and it downloads....another 9201 WU. Same result. How can i prevent to waste time and ressources for those bugged project 9201 WUs?

[bollix47]

Next time it gets 'stuck' simply Pause any running slots, restart the client and don't delete anything. If restarting the client doesn't work then add a reboot between the Pause and the restart/Fold. Also, ensure you're using the latest client 7.4.4.

[toTOW]

We need to see the log from the client to understand where it is getting stuck ...

[ThePegasus1979]

Next time it gets 'stuck' simply Pause any running slots, restart the client and don't delete anything. If restarting the client doesn't work then add a reboot between the Pause and the restart/Fold. Also, ensure you're using the latest client 7.4.4.

That didn't help repeatingly. I now just deleted the work folder over and over again (~12 times) until i got a different WU. Client is 7.4.4.

@ToTOW

There was absolutely nothing useful in the logs, even with verbosity set to "5".

[toTOW]

We need to see the logs anyway if you want us to help ...

[bollix47]

Also, please don't delete the work folder as doing so means the server will not get what information is available.

Although that way was recommended in the past, the more efficient and some would say correct way to dump a work unit is to pause the slot, then go into Configure > Slots and Remove the offensive slot. After a few seconds you can Add it back and click on Fold if necessary. This way the client will sent a message to the server that the WU was dumped, set it up for reassignment and send you a new WU.

Otherwise, if the server doesn't know what you've done it will wait until the timeout date to reassign it or may even send it to you again.

A restart of the client has worked for most people when they've run into this problem but some have found they needed to also reboot their computer and/or any communication devices on their lan such as routers or modems.

[ThePegasus1979]

Also, please don't delete the work folder as doing so means the server will not get what information is available.

Although that way was recommended in the past, the more efficient and some would say correct way to dump a work unit is to pause the slot, then go into Configure > Slots and Remove the offensive slot. After a few seconds you can Add it back and click on Fold if necessary. This way the client will sent a message to the server that the WU was dumped, set it up for reassignment and send you a new WU.

Otherwise, if the server doesn't know what you've done it will wait until the timeout date to reassign it or may even send it to you again.

Oh, ok, sorry. Will stick to that advice if it happens again, what i don't hope to happen

[bollix47]

In case you didn't notice I added an edit to my above post before I saw you had responded:

A restart of the client has worked for most people when they've run into this problem but some have found they needed to also reboot their computer and/or any communication devices on their lan such as routers or modems.

[ThePegasus1979]

In case you didn't notice I added an edit to my above post before I saw you had responded:

A restart of the client has worked for most people when they've run into this problem but some have found they needed to also reboot their computer and/or any communication devices on their lan such as routers or modems.

Thanks. As the CPU folding slot sent and recieved successfully during that 9201 issues, i doubt it was related communications issue. Especially as getting new WUs for the GPU after deleting the corresponding "work" folder wasn't a problem either, and a couple of FAH as well as system restarts didn't solve the 9201 getting stuck at 99.99% issue either.

[Joe_H]

The client showing GPU WU's going to 99.99% but not actually completing has usually been connected to a driver reset in the middle of processing. Checking the log file would show that the WU reached a lower level of completion, and then no for progress. The client keeps running the estimate after losing communication with the folding core.

As has been already mentioned, restarting the FAHClient process, not just FAHControl or the web control, usually will complete one of these stuck WU's. Post your log file if this happens again, and we can confirm this happening for you. Causes for a video driver reset can be overheating, excessive overclocking of a GPU, or using the Remote Desktop to access the PC. Examination of your Windows log for the time that corresponds to the WU no longer showing progress in the log file can also confirm the driver reset.

[bruce]

There was absolutely nothing useful in the logs, even with verbosity set to "5".

We need to see the logs anyway if you want us to help ...

If it happens again, do post the log, and avoid posting anything with elevated verbosity.

The following may help, too:
* Find the last line containing HH:MM:SS:...:Completed N out of M steps (80%) for the applicable WU.
* Convert HH:MM:SS to your local timezone.
* Open the Windows "Event Viewer" and you should find an error report from the GPU.