any rough idea when it will work again?

Yes - IBM Spectrum Scale FPO (i.e. shared nothing). IBM Spectrum Scale is the new marketing name for GPFS.

We really like it also. We have been talking to level 3 support and it appears that there are some configuration options set wrong that have made these events much more likely to occur. They have given us a disk check command that is running MUCH faster (to give you an idea, before we were talking to them the check was taking about 3.5 hours per 1% scanned). We restarted it based on their recommendations and it is now at 65% complete so we are hopeful to be back online in the next couple of hours.

We will also be working to fix the configuration options that should give us better stability for this cluster.

All of this means that - yes this is related to the move, but no it isn't related to the cloud.

We are on a new version and using the shared nothing options so we are encountering some new things that once we get them resolved, we expect this to become a distant memory (well, at least distant once some distance has occurred)

+1!

Not only is it possible,but also a notice WAS sent to each machine which could be read in your BOINC manager notices World Community Grid: Short Planned Outage for Tuesday, July 18

The techies do a great job.

Give them time.