This also coincides with an outage I've noted today at Rosetta@home.

Rosetta is based in the U.S. (Seattle, Washington).
They just have a shortage of work at the moment. Their users have exploded four times since the virus started.

Suspending to try and fix the full disk.

The server's disk is just a buffer. It is emptied continuously. Disk full conditions happen when there is even a temporary imbalance between in (uploads) and out (moving to the main servers) rates. A few hours of imbalance are sufficient to fill it. At such high volumes there is no "easy fix".

thanks for explaining.

After uploads worked again for a few hours this late morning, they have stopped once more since some time ago :-(

Oh. I had suffer from it. It just started and turn to "Try again later" immediately in few seconds. Ohh... Please! fix it, quick is better.

The server's disk is just a buffer. It is emptied continuously. Disk full conditions happen when there is even a temporary imbalance between in (uploads) and out (moving to the main servers) rates. A few hours of imbalance are sufficient to fill it. At such high volumes there is no "easy fix".

Now I understand the 2 WU per GPU limitation, you're trying to balance the goes inners and the goes outters.

At such high volumes there is no "easy fix".

a larger disk buffer?

---

Having the same issue. 6 GPU tasks refusing to upload.

*Edit: Ah, they just went through...

This is still a big problem. I have to retry to submit completed WUs several times a day to get idle GPUs working again. Please fix server issue.

If you can't fix your server problems you could at least let us download a day's worth of WUs. That would be a minimum 12 WUs per GPU.

This is still a big problem.

I haven't seen it, though I am running only two GTX 1060s at the moment. Since your computers are hidden, there is not much more to say.

Every morning I wake up to most of my GPUs sitting idle waiting for GG to UL & subsequently DL WUs. I know of no other project that is any where near as inefficient at keeping work supplied.
Why don't you get someone from another project to look at your server configuration??? Maybe valterc at TN-GRID, the most efficient BOINC project going.

I'm not having any issue with uploads or downloads.

---

I have also frequent upload problems. Not always, not across all my systems, I haven't found a common way of ocurring but it is really happening since many months ago. And the access to the project website is very slow most of the time.

I'm using ubuntu in all my hosts.

After about 3 hours all GPUs have gone idle because WUs do not UL and so the paltry 2 WUs do not DL.

After about 3 hours all GPUs have gone idle because WUs do not UL and so the paltry 2 WUs do not DL.

There is a known issue on GPUGrid that after you access the server, there is some dead time before you can access it again. I am not seeing it at the moment, but it bites everyone eventually. That may be it, depending on how many machines you have.

I would guess that it is some sort of anti-DDOS protection feature on the campus network, but no one knows (or admits to) what the problem is.

There is a known issue on GPUGrid that after you access the server, there is some dead time before you can access it again...

This problem was discussed at the end of 2019 on thread Unable to load units
Currently I have 7 hosts in production, all of them attached to the same local network.
I've tested that (for me), when I want to replenish WU buffers, the most effective way is manually asking for WUs one by one host, from lowest to highest local IP. I've configured fixed IP address for each one.
These 7 hosts manage 11 GPUs in total (some of them are multiGPU systems), so the maximum WUs I can aspire to simultaneously download from GPUGrid is 22 (two at a time per each GPU)...
But most of the time, the whole group is quite well automanaging by means of their individual BOINC Managers, without human intervention.

no finished task can be upoloaded.

19.12.2020 08:42:24 | GPUGRID | [error] Error reported by file upload server: Server is out of disk space

Same here:

19.12.2020 11:45:10 | GPUGRID | Started upload of 3m0eA01_379_2-TONI_MDADex2sm-43-50-RND8091_0_10
19.12.2020 11:45:11 | GPUGRID | [error] Error reported by file upload server: can't open file /home/ps3grid/projects/PS3GRID/upload/96/3m0eA01_379_2-TONI_MDADex2sm-43-50-RND8091_0_10: No space left on device
19.12.2020 11:45:11 | GPUGRID | Temporarily failed upload of 3m0eA01_379_2-TONI_MDADex2sm-43-50-RND8091_0_10: transient upload error

Today is Saturday ... I hope they solve the problem next week ...
K.

---

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

Today is Saturday ... I hope they solve the problem next week ...

it's not the first time this happens; so I am surprised that they havn't installed any warning system to the effect that someone gets a notification as soon as the disk is about 80% full or so.

Still down: 12/19/2020 1:13:30 PM (CET) | GPUGRID | [error] Error reported by file upload server: Server is out of disk space