Sometimes machines hang for no apparent reason (although there is a reason which may be hard to track down.) It is a good policy to just reboot them from time to time. I am guessing that if you reboot the machine in question it will go back to processing units just as before.
Cheers

jmcgaw said:

Ha!
The ability to merge duplicate device/host IDs into one has been a requested feature on WCG for years but hasn't been a priority for them. Other BOINC projects [that don't use custom or legacy code on the back-end] have that feature by default. I have a duplicate pair I'd like to merge as well. I hope this functionality becomes a reality on WCG. I'm holding my breath.

Regarding your "No Reply" question, I bet the work units that show "No Reply" all correspond to the original "Beauty" device that last returned results on 10/2/19. Have you changed any hardware on Beauty that would cause WCG server to reject it and create the new Beauty device? In other words, have you upgraded RAM, installed a new hard drive or SSD, or updated to a newer Windows 10 version?

jmcgaw said:

That would do it. There's a "client_state.xml" file in C:\ProgramData\BOINC that has some kind of counter that gets incremented each time the BOINC client talks to the server, and if it detects that the counter is lower, it'll abandon ship and create a new device/host ID. So when you restored an image that was two days older, it restored an older client_state.xml file with a previously used counter. Any work units in progress would simply never be heard from again since they were deleted (which explains the "No Reply" status).

I think to avoid this in the future if there's ever a need to re-image the computer, make sure BOINC is completely uninstalled and C:\ProgramData\BOINC folder deleted then BOINC reinstalled before the machine even connects to the Internet.

In the meantime, join me in holding our breath for the ability to merge devices. :P

That is something I should probably look into although this is the first time (that I recall) that anything like this has happened to me. It might prove to be a problem on a few of my stripped-down "crunching-only" machines which have only a single disk and partition remaining. Or maybe I'll just forget about the whole thing and not worry about lost credit -- I'm not competing with anybody but myself and as long as the units get crunched then the science is being done and that is the only real reason I spend my $$$ on electricity toward the projects.

Then again, I wonder if the boinc data could be stored on one of the redundant Drobo NAS units?...

@hchc
I do systematically apply some segregation principles by configuring systems.

For Windows-based systems, e.g.:
> C: System only: OS + Applications, without any data
> D: Data, including boincdata directory
> E: Documentation material, installation directories ($setup$.exe, .msi, ...)
> F: Entertainment, e.g. .mp3, .mp4, ...
This strategy allows to differentiate the priority of backup activities, to generate images (C:) without any data, just system stuff, etc.

On Unix/Linux based systems, I follow the recommended Standard Linux File System Structure. Excepted for VM without much data, I configure systematically the machine with a separate /home partition. It's allow me to be able to reinstall "from scratch" a machine without jeopardizing my data, taking care that I do never allow the installation script to partition the disks on its own.
In this case, the boinc data are stored in /var/lib/boinc-client.
You need to be careful by reinstalling a system, since, even if /var resides on a dedicated partition, the /var directory structure will be fully resetted during a new system installation (I made the failure on day and loose my boincdata directory).
For avoiding this problem, I stop the boinc project using boinc manager:
1/ no new work
2/ suspend project (as soon as all WUs have been processed)
3/ Shutdown boinc).
Afterwards I copy /var/lib/boinc-client to home.
After the new system installation, I copy boinc-client back to /var/lib before installing boinc on the new configuration. Normally, assuming that you took properly care of ownership, group membership, and access rights, the machine should be immediately operational after the boinc installation. Afterwards, you should not forget to reactivate the project using boinc manager:
1/ Resume project
2/ Allow new tasks.
Hopefully, the explanation is clear enough.
Happy crunching,
Yves
---
PS: Regarding Windows, I elaborated this segregation strategy already during the MS-DOS era (>30 years ago)

PPS: After different ideas for Unix/Linux systems, I create an installation repository in /home/sw-depot for the software staff I do manually install. Likewise if the machine has to operate virtual machines (VM), I create a dedicated partition - /VM - for storing the virtual disks (.vdi, .vdmk).