Core 21 0.0.18 - Hangs on core update

[DocJonz]

I've had a few GPU WUs upload within the last minutes, and then they hang waiting for the new core update. When the message from The Official Blog says the new core is being 'pushed out', it needs to be pushed out a bit harder so as not to hang clients(!)

[bollix47]

Try removing your old core.

[Joe_H]

Are these GPU's in the same system? If so, the update is waiting on all folding slots using the old version of the core to complete. The client is not designed such that the core can be updated while it is actively being used. That is common to all software, I know of few programs that can do updates while active and those are usually in the critical availability field of computing.

[DocJonz]

Are these GPU's in the same system? If so, the update is waiting on all folding slots using the old version of the core to complete. The client is not designed such that the core can be updated while it is actively being used. That is common to all software, I know of few programs that can do updates while active and those are usually in the critical availability field of computing.

They are all dual GPU systems, so that sounds like a probable cause. Guess I have to be more patient and let the other WUs complete(!)

[davidcoton]

You may still need to delete old core21s manually and/or reboot -- I needed some of that even on a single GPU Linux box.

[bruce]

The trick is that if a WU hangs waiting to install a new FAHCore, go to any other slot that is running the same FAHCore and PAUSE it. Once the new core is installed, the slot that was hung will start processing and you can UNPAUSE (akd "FOLD") the slot(s) that you paused. [Slots that are running a different FAHCore need not be paused.\

[Joe_H]

Someone also reported in the past that the update on a multi-GPU system often went better if the slots still active were set to Finish. That would keep more than one update from happening simultaneously.

[road-runner]

I had a dual gpu machine doing the same thing while ago, I rebooted and its fine or seems to be

[bruce]

Reboot won't always work, though it may.
Pausing the "other" slot(s) using the same core should.
So will removing the old core.