If forced to, how should I manually dump?

[Knish]

a work unit, that is. Normally I would wait until all WUs get returned before decomissioning the PC, but what if I wrongly assume I have enough time for 1 more WU and it turns out to be a monster with a 15 day timeout. I'd rather not lock up research for that long if I know that rig can never return to folding. Due to accidents, I know some user-fault dumps get uploaded to the server so they know to immediately re-issue the WU, but I have no idea what the exact steps are.

[ajm]

You have to pause the slot, then delete the corresponding work folder. In a normal Windows installation, you'll find it there: %AppData%\FAHClient\work
If you have only one slot, you can delete the whole work folder. Otherwise only the folder 00, 01, 02, or 0n corresponding to the ID in FAHControl.

[Knish]

for some reason I was under the impression that doing it that way doesn't tell the server of the dumping. Are you sure the Dumping gets communicated back to the server that way?

[ajm]

In one word, no. It's just a way not to waste your ressources. I don't know of any way to specifically "tell" the server that you dumped a WU.

[Knish]

During my pre-4tho'July debacle i found out aparently denying a GPUs.txt update will cause the core_22 to dump its WU and tell the server. But now I actually want to know the method for both cores, cpu & gpu.

edit> oh, and I should mention I'm interested in Linux, and I know FAHClient --send-command dump ____? is a possibility (or is that FAHClient --dump?) but can't figure out what the help doc is telling me, nor do I want to just willy nilly try things

[bruce]

If you're planning on shutting down anyway, why bother. You can't inform the server you've dumped, so don't dump -- just shut down. The WU will eventually expire and the server will know.

[Knish]

as an FYI i have noticed at least twice that an upload has followed a Dump, so the server was able to reissue within a few minutes. I bring this up b/c today I noticed one of the antartica bugs has a Timeout of 15 days, that would suck if I end up getting that again on shutdown day

[bruce]

FAHClient does make an effort to report dumped WUs. It can't always do that.

[gunnarre]

Just delete all folding slots while the client is running. The client will report the work unit as dumped if it has no slots to process it on.

If you have FAHControl, just click Configure, choose the Slots tab, Click Remove on all the slots until the list is empty, then save, see that the slots disappeared and wait for the client to connect to a work server and report all the work units as dumped. You'll see that it's done with that when the Work Queue is empty.

Code: Select all

10:03:31:WARNING:WU01:Slot ID 0 no longer exists and there are no other matching slots, dumping
10:03:32:WU01:Sending unit results: id:01 state:SEND error:DUMPED project:13851 run:0 clone:30215 gen:220 core:0xa7 unit:0x00000112287234c95e78846c4be6296b
10:03:33:WU01:Connecting to 40.114.52.201:8080
10:03:34:WU01:Server responded WORK_ACK (400)

I don't dump work units unless I have to, like if I upgraded the kernel and OpenCL no longer works with my oldest graphics card.

PS: Do not edit the GPUs.txt file yourself - that counts as hacking the FoldingAtHome software, and it's against the rules as far as I can tell.

[bruce]

PS: Do not edit the GPUs.txt file yourself - that counts as hacking the FoldingAtHome software, and it's against the rules as far as I can tell.

And it's useless. The GPUs.txt is cached on your HD to make it easier for FAHControl to configure your GPUs, but when your GPU connects to the servers the entry is validated against the master copy on the server, itself. Strange things happen if you try to download a WU that's assigned to an unsupported GPU.