Uploads are stalled out again.

this time, obviously not "disk full", but "transient upload error"

This is what I am getting in the event log:

Sat 22 Jun 2024 01:15:26 AM EDT | GPUGRID | Started upload of cdk8_m40_m19_5-QUICO_ATM_GAFF2_RESP-2-7-RND1453_1_0
Sat 22 Jun 2024 01:15:44 AM EDT | GPUGRID | Started upload of cdk8_m40_m19_5-QUICO_ATM_GAFF2_RESP-2-7-RND1453_1_1
Sat 22 Jun 2024 01:15:45 AM EDT | GPUGRID | [error] Error reported by file upload server: Server is out of disk space
Sat 22 Jun 2024 01:15:45 AM EDT | GPUGRID | Temporarily failed upload of cdk8_m40_m19_5-QUICO_ATM_GAFF2_RESP-2-7-RND1453_1_1: transient upload error
Sat 22 Jun 2024 01:15:45 AM EDT | GPUGRID | Backing off 04:47:11 on upload of cdk8_m40_m19_5-QUICO_ATM_GAFF2_RESP-2-7-RND1453_1_1
Sat 22 Jun 2024 01:16:00 AM EDT | GPUGRID | [error] Error reported by file upload server: Server is out of disk space
Sat 22 Jun 2024 01:16:00 AM EDT | GPUGRID | Temporarily failed upload of cdk8_m40_m19_5-QUICO_ATM_GAFF2_RESP-2-7-RND1453_1_0: transient upload error
Sat 22 Jun 2024 01:16:00 AM EDT | GPUGRID | Backing off 03:39:30 on upload of cdk8_m40_m19_5-QUICO_ATM_GAFF2_RESP-2-7-RND1453_1_0

now I get the same:

[error] Error reported by file upload server: Server is out of disk space

they had this problem day before yesterday, it was then solved yesterday - and now they have it again ?

Hi, same here ... disk out of space ...
K.

---

Dreams do not always come true. But not because they are too big or impossible. Why did we stop believing.
(Martin Luther King)

I've seen this before, at another project that produces very large upload files.

The upload server can't store the total volume of incoming data by itself, so it stages the files temporarily, and then passes them on to a larger, but slower, mass storage device.

It can happen that the upload server is full of staged files (new requests are rejected immediately), or it can accept them for upload, but find that the onward link can't accept them for transfer to backing storage (the file appears to upload up to 100%, but is then rejected).

I hit both versions of the problem about an hour ago, but trying them again just now got them through. So, the occasional retry may be worthwhile, but please don't keep hammering the retry button - that'll just make things worse for everyone.

the problem is back again since late afternoon :-(

the problem is back again since late afternoon :-(

the uploads worked some time last night, but now again they don't.

What's going on over there?

Same problem disk full ... it's sunday ... I'll see you on monday, i hope
K.

---

Dreams do not always come true. But not because they are too big or impossible. Why did we stop believing.
(Martin Luther King)

Up to 6 new app versions just out of the oven today, 26 Jun 2024.
For both Linux and Windows, CUDA1121 every of them.

Up to 6 new app versions just out of the oven today, 26 Jun 2024.
For both Linux and Windows, CUDA1121 every of them.

Just noticed that on my latest ATM download - it's v1.16, instead of v1.15

The most obvious change is that my BOINC client has estimated the task at nigh on 3 days, compared with 7 hours for the previous batch. I'm putting that down to the usual quirks of runtime estimation at this project, but I'll keep an eye on it.

https://www.gpugrid.net/result.php?resultid=35373201

https://www.gpugrid.net/result.php?resultid=35373196

And all of them are crashing on my computer ...

Bevore this only the ATM crashed ...
This is rly bad for me in case i like GPU Grid rly but i can not find the problem ...

http://www.gpugrid.net/forum_thread.php?id=5454#61506

---

This wont fix the Problem in my case ...

I tried 40GB and 30GB and all WUs are crashing ...

try by disabling the memory integrity on windows 11 ,after these are the only 2 problems that I had and that prevented boinc from working properly on my pc windows.

essayer en désactivant l'intégrité de la mémoire sur windows 11 ,apres ce sont les 2 seuls problemes que j'ai eu et qui empechaient boinc de fonctionner correctement sur mon pc windows

---

There is a new app for windows to run ATMML and ATMbeta.

There is a new app for windows to run ATMML and ATMbeta.

Right.
Current app versions for each OS can be checked at GPUGRID apps page.
Congrats for Windows users.
And Well Done! for project developers.

There is a new app for windows to run ATMML and ATMbeta.

Right.
Current app versions for each OS can be checked at GPUGRID apps page.
Congrats for Windows users.
And Well Done! for project developers.

+ 1

I am using gpugrid for some years. With ATM: Free energy calculations of protein-ligand binding v1.19 (cuda1121)
x86_64-pc-linux-gnu i've got a problem.

The workunit runs quite some days, roundabout 45 to 55 % a day. But every morning i restart my PC and Boinc (7.24.1) the workunit starts again at 0 %.
I am using Ubuntu 24.04.2, CPU AMD Ryzen 7 5800 X x 16; GPU NVIDIA GeForce GT 1030

What can I do?

Don't restart your computer while an ATM task is in progress.
The ATM tasks does not make checkpoints, so they start at 0 if something interrupts them.
This is a known bug. Hopefully it get fixed, as it is very annoying.

It was fixed. But unfortunately, with the new app after the server migration, that piece of code has disappeared again.

I've asked the question in the Discord if they would consider reenabling that, but no reply yet.