HTTP_SERVICE_UNAVAILABLE

[Greatnessguru]

01:04:58:I1:WU352:Requesting WU assignment for user Anonymous team 0
01:04:58:I1:OUT22:> POST https://assign2.foldingathome.org/api/assign HTTP/1.1
01:04:58:I1:OUT22:< HTTP/1.1 503 HTTP_SERVICE_UNAVAILABLE
01:04:58:E :OUT22:HTTP_SERVICE_UNAVAILABLE: {"error":"No appropriate assignment"}
01:04:58:I1:WU352:Retry #20 in 8 mins 32 secs

This error msg repeats every 8 mins 32 secs.
I have no running WUs.
Can someone fix the server, pls?
The assign2 link above says: error "Key 'data' not found"

[Joe_H]

The "No appropriate assignment" means there are no WUs available suitable for your system with the hardware available and client settings you have made. If this is your Raspberry Pi 5, that could mean that all of the WUs from the smaller projects are assigned or they are currently inactive.

[Greatnessguru]

OK. Thank you.

[Paul_M]

I hope it's OK me ju,ping on here rather than starting a new thread.
I've got the same issue in that I haven't received a WU for at least 7 days and I get the 'no appropriate assignment' message.
I'm running on a mac - are there no assignments available for mac hardware, or is something messed up in my settings?
Thanks

[Joe_H]

Well, it is not because you are running on a Mac, if that was the case I wouldn't be getting any WUs either. There is currently a 3 CPU thread minimum setting on most CPU projects though. So if your settings are for 2 threads, it is possible to not get any work at the current time.

If that is not the case, could you post the beginning section of your log file that shows the system setup and a section where a WU request is not filled.

[Paul_M]

Ah - bingo. I'm only running 2 cores. I'll take it up to 3 and see what happens.
Thanks

edit - straight away a WU starts up. I wish I'd known that sooner ...

[Joe_H]

Hopefully that 3 thread minimum will go away at some point, it is a workaround for a problem that can show up with the v7 client. If the client is configured wrong it tries to load a CPU WU instead of a GPU WU.

[toTOW]

No, the issue is with v8 default resource group that allow to mix GPU and CPU allocations. v7 client is perfectly fine as CPU and GPU slots are exclusive.

[arisu]

No, the issue is with v8 default resource group that allow to mix GPU and CPU allocations. v7 client is perfectly fine as CPU and GPU slots are exclusive.

Why not allow <=2 threads if the client is making a request that doesn't indicate a GPU in the JSON sent to /api/assign? That would solve it for everyone who is doing CPU-only folding, which is probably almost everyone who wants to fold on so few threads anyway.

Example function (in Python) that would do this server-side:

Code: Select all

def should_assign(request: dict) -> bool:
    resources = request['resources']
    if resources['cpus']['cpu'] < 3:
        for resource in resources.keys():
            if resource.startswith('gpu'):
                return False
    return True

[muziqaz]

No, the issue is with v8 default resource group that allow to mix GPU and CPU allocations. v7 client is perfectly fine as CPU and GPU slots are exclusive.

Why not allow <=2 threads if the client is making a request that doesn't indicate a GPU in the JSON sent to /api/assign? That would solve it for everyone who is doing CPU-only folding, which is probably almost everyone who wants to fold on so few threads anyway.

Example function (in Python) that would do this server-side:

Code: Select all

def should_assign(request: dict) -> bool:
    resources = request['resources']
    if resources['cpus']['cpu'] < 3:
        for resource in resources.keys():
            if resource.startswith('gpu'):
                return False
    return True

There are many ways of constraining the assignments. 99% of them are not simple and/or easy

[arisu]

That would fix it and would be simple though. Or is the server-side code too convoluted to make a change like that?

[muziqaz]

That would fix it and would be simple though. Or is the server-side code too convoluted to make a change like that?

Server code is not us here, but the person who has too many responsibilities

[Greatnessguru]

I'm getting WUs again. I'm using all four cores.
Thanks.