Project URL out of date

 You should have just tested it with an Einstein account file and left poor old Primegrid alone :-).

 I'm pretty sure that MarkJ suggested a new state file (which would have been created from information sent back by the scheduler) because (like me) he didn't realise that Primegrid was already running and had already created a state file :-).

 That's great news.  I'm really glad you have got it sorted.

This is why when I refurbish an old machine, I always create a quite small template state file with just a couple of crucial bits of information that guarantee the machine will use the former ID, even if there has been a complete change of OS and hardware.  It's easy for me since I only support one project.

With multiple projects, it's still possible.  You use the existing state file and add a new <project> ... </project> block to it.  That added block would be quite short with basically just the critical information.  The server just needs to know that the old host ID does indeed belong to the account listed in the account file and then it will fill in for you all the vast quantity of lines of information that normally reside in a specific project block.  Once you've done it a couple of times, it's a pretty simple thing to do.

Here is a classic example of where this has proved useful to me.  About 2 weeks ago, a late summer storm caused a fraction of a second power glitch which took out 90% of the fleet.  A couple of machines had severe file system corruption and wouldn't reboot.

In those or similar situations, the easiest recovery is to install a replacement disk, install the OS from a live USB, use a script to install all the BOINC stuff from templates, modify the template state file to give the correct host ID, launch BOINC and have it contact the project to get the 100's of tasks that the machine had at the time of the storm, resent as 'lost tasks'.  In less than an hour, the damaged machine was back crunching those very same tasks, with no other volunteer having to wait two weeks for those otherwise inaccessible tasks to time out.

The corrupted disks that were removed have been completely zeroed, re-partitioned, re-formatted and read/write checked for bad blocks and put in the stockpile for the next time this sort of event happens :-).  I have quite a collection of old disks (many 20GB IDE) that were put into service in business machines around 2002.   I purchased (and put back into service as crunchers) those machines in lots of 10 in computer auctions in 2006/7 when they were listed for sale.  I ran them for a couple of years and then shut down most of them and upgraded the rest with new board/CPU/RAM around 2009.  I have a big stockpile of spares to choose from :-).  I'm still using the disks and some of the PSUs.  Amazingly, very few of the disks have ever failed.