I haven't been able to send completed work units for a couple weeks now. Even though all the servers are green across the board, I get:
Project communication failed: attempting access to reference site
Access to reference site succeeded - project servers may be temporarily down.
I have been sending/recieving Seti work units all along, so I'm confused where the issue could be. About the only change is a CPU upgrade on my end, but that shouldn't matter for just sending the work units. . .
Copyright © 2025 Einstein@Home. All rights reserved.
Not sending work
)
You have more than one PC under your account, at least one shows normal operation. Can you point us to the host that is having troubles?
Anyway, this sounds like a firewall problem. The messages you quoted mean that BOINC could not reach a site of the Einstein@Home project, but was able to contact other internet sites used for testing internet connection.
It could be that this particular PC is behind a firewall that either explicitly limits access to some BOINC sites (but then SETI should be blocked as well) or has some rules that happen to block E@H file uploads.
Is this a PC connected directly to the internet or is it behind a corporate firewall?
CU
Bikeman
RE: About the only change
)
Was this done during you having work? What was the old CPU and what is the new one?
A CPU change while you have work can account for BOINC seeing it as a different PC, making it unable to upload work that isn't registered to that computer. But I'm not sure what error message you should get then.
The PC is 951391 and the
)
The PC is 951391 and the upgrade was from a Dual Core to a Quad. It is behind a firewall, but it should not be blocking the traffic. It is also requesting quite a lot of work (892283 seconds) that doesn't get fulfilled. I suppose I could always re-register the box, but I'd hate to lose 120+ hours of crunching.
The last time the PC with
)
The last time the PC with that ID made contact with the EAH scheduler was Oct. 26th, so to me it looks like it really can't get thru.
Whatever you do, you will not loose the credits for past crunching, at worst it will be listed for a computer that is no longer active, but the credits will stay with your personal "account".
You could do the following:
"Reset" the project. That means that all downloaded files will be refreshed (deleted and downloaded again) and some internal stuff is "reset".
If that doesn't help: detach the PC and re-attach under a new host-id.
You might even be able to merge the two virtual copies of your PC again.
Anyway, changing the number of CPUs on-the-fly might well confuse BOINC and it might be a good idea to at least reset anyways.
Hope this helps,
Bikeman
A reset seems to have cleared
)
A reset seems to have cleared it up. (Painful to lose 120+ hours of work completed though)
Thanks!
It looks like I spoke too
)
It looks like I spoke too soon. It downloads tasks just fine and looks to be communicating, but it still won't upload completed tasks.
RE: It looks like I spoke
)
I bet it's a firewall or proxy issue. Maybe for some strange reasons it's configured to disallow POST request exceeding a certain size. E@H is using HTTP POST to upload files which in itself should be allowed by any firewall, but the payload is ca 70 ...80 kB and this might be a problem.
CU
H-B