I have three Ubuntu 20.04 servers that have been members of the Rosetta project form some time. Recently with the python work units, Rosetta will disregard my BOINC client drive space settings and the project level "No New Work" setting. My BOINC clients on these three hosts are set to provide only 45GB of space for BOINC. The project will fetch work and fill my drive, sometimes consuming in excess of +65GB of space. When Rosetta does this, the BOINC client shuts down. I have to restart it, abort a number of regular and python tasks, clear out the /slots/ directory, and let the client resume computation with the other projects. I do not have an app_config/xml for the Rosetta project as I've had too may issues with the max concurrent settings I used to have.

Is anyone else experiencing this? I've had to remove the Rosetta project to keep the hosts online.

You haven't re-enabled max_concurrent have you?
Even so, setting "No new tasks" should stop BOINC from requesting new work altogether, regardless of max_concurrent usage,


Might be worth a post to the BOINC forums.

---

Grant
Darwin NT

You haven't re-enabled max_concurrent have you?

No, I haven't. I don't have an app_config for Rosetta at all. I do have them for LHC@Home and GPUGrid, but their work fetch is working just fine.
I'll go ahead and post it to the BOINC forum and go from there.

I have had success with this version:
<project_max_concurrent>n</project_max_concurrent>
But that is project level and will only allow however many you use.
It does not see the difference between 4.2 or Python.