Hi,

Now that GPUGrid is finally back, I'm getting this error:

2/25/2025 12:06:11 PM | GPUGRID | Scheduler request to https://www.gpugrid.net/PS3GRID_cgi/cgi failed: HTTP file not found

Anyone know why or what this means or how to fix it?

Thanks.

Doug
[/url]

Hi, yes the server is back but various things have changed, including some of these internal urls/paths. We are still testing.
- If you re-attach to the project these should be fixed (they are set by the get_project_config.php)
- I will look into setting up some redirects from the old paths to the new ones

I got caught by another variation of the same problem. My messages when the project came back up were:

25/02/2025 17:32:12 | gpugrid | Scheduler request completed
25/02/2025 17:32:12 | gpugrid | [sched_op] Server version 715
25/02/2025 17:32:12 | gpugrid | Master URL changed from https://www.gpugrid.net/ to https://gpugrid.net/gpugrid/
25/02/2025 17:32:12 | gpugrid | Resetting project
25/02/2025 17:32:13 | gpugrid | Project requested delay of 11 seconds
25/02/2025 17:32:13 | gpugrid | [error] Server requested conditional abort of unknown task hif2a_A40_A37_r0_2-QUICO_ATM_AF_04_Benchmark_hif2a_v3-9-10-RND0712_0
25/02/2025 17:32:13 | gpugrid | [sched_op] Deferring communication for 00:00:11
25/02/2025 17:32:13 | gpugrid | [sched_op] Reason: requested by project
25/02/2025 18:01:34 | gpugrid | work fetch resumed by user
25/02/2025 18:01:37 | gpugrid | [sched_op] Starting scheduler request
25/02/2025 18:01:37 | gpugrid | [sched_op] Fetching master file
25/02/2025 18:01:38 | gpugrid | [sched_op] Got master file; parsing
25/02/2025 18:01:38 | gpugrid | [sched_op] Found 1 scheduler URLs in master file
25/02/2025 18:01:38 | gpugrid | Master file download succeeded
25/02/2025 18:01:43 | gpugrid | [sched_op] Starting scheduler request
25/02/2025 18:01:43 | gpugrid | Sending scheduler request: To fetch work.

and the rest is normal.

I think task hif2a_A40_A37_r0_2-QUICO_ATM_AF_04_Benchmark_hif2a_v3-9-10-RND0712_0 must be the one I reported in ATMML, which was issued (as a new task) just before the shutdown, but was denied access to report it 12 hours later. (*)

This is a security feature of BOINC, and has been for 20 years - it makes it much harder for interlopers to take control of the url and to steal the project's scientific results. But it always surprises me when project administrators don't seem to be aware of it.

Don't feel too bad about it - CPDN got caught out in the same trap last year. And at least I get a whole lot of free disk space back ("Resetting project" deletes everything).

But watch out for the download bandwidth spike when you issue the first new batch of work!

(*) - yes, confirmed

Hi,

Now that GPUGrid is finally back, I'm getting this error:

2/25/2025 12:06:11 PM | GPUGRID | Scheduler request to https://www.gpugrid.net/PS3GRID_cgi/cgi failed: HTTP file not found

Anyone know why or what this means or how to fix it?

Thanks.

Doug
[/url]

I just clicked on the reset button in BOINC manager.
The click on update
You will get this message: Master URL changed from https://www.gpugrid.net/ to https://gpugrid.net/gpugrid/
It will download the master file and then pause.

Hi,

Thanks for the info. I followed your procedure and the error is gone. Alas, I didn't get any tasks (the Status page says there's 3, but I'm not sure how reliable that is).

Thanks again.

Doug

Hi,

Thanks for the info. I followed your procedure and the error is gone. Alas, I didn't get any tasks (the Status page says there's 3, but I'm not sure how reliable that is).

Thanks again.

Doug

I saw those last night, but I guess they are stuck in the system.
ATMML: Free energy with neural networks 3

Isn't that what we were working on before the server move?

yes those 3 are old and stuck

Hi, yes the server is back but various things have changed, including some of these internal urls/paths. We are still testing.

I see, in the server status page, that the project is using 1.2.0 version of the Boinc server (May 2020)
The latest is 1.4.5 (December 2024). Do you plan to update?

Four brand-new ACEMD 3 test tasks downloaded, crunched, and reported. No problems encountered.

(except for the predicted slow re-download of the application packages: three identical 1.3 GB files over a shared internet link.)

Steve - why are there 27 tasks that are sitting in the queue but not being sent out?
Are they all stuck and you haven't removed them or what is going on?
When do we get new tasks?

Asked in the Discord page and got this answer:

Steve is on leave until next week.
For example, in ATM and ATMML, the apps need to be updated to what we have locally. Once this is done I have loads of work to send
We have made some changes, mainly performance/speed improvements

Steve himself posted a little more detail in message 62242 on this board.