The way I would approach this would be:

1. Stop and restart the client. Then look in the boinc client log for the number of CPUs the client has detected. It should be 80.

2. if not, set the NCPUS value in cc_config.xml to 80 and re-read the config files. Look in the log for a message that it was set to 80 CPUs.

3.If the client has 80 CPUs set and you are getting the limit of tasks in progress message. My first thought would be that you are hitting the 1000 WU per host hard limit. The BOINC client can only have 1000 WUs in progress regardless of CPUs or work queue length. This is a BOINC restriction not a project restriction.

If you don't have 1000 WUs assigned to that host and are getting a limit message, I would next check to see if you are getting errors on that machine which would restrict the number of WUs you get until you send enough valids to raise the limit. If not getting errors, then I would send an email to support@www.worldcommunitygrid.org (I think that is the address) for help

Sorry for not replying more quickly. Had a few things to sort out.

Ok. Just to explain the picture a little further.
Excluding android phones, I currently run 14 computers, almost all 24/7 and have always run them with buffers. Some just run with the global preferences, some I tweak at the local level depending on their situation.

Most of them have been running for years with no issues at all, which is how I have almost 1000 years of run time now.

Suddenly, 3 months ago, all my computers started to run without buffers. At most they have 2 tasks ready to run, when I have two tasks about to finish. In reality, it is a 0 or a 1 task buffer.
Of the 14 computers, 11 run windows and 3 run linux.

At the moment, I have about 428 tasks in pending validation/verification, 0 invalid, 0 too late, 0 aborted. There are 4 tasks with errors on 2 machines. Over the last 31 days, I have averaged 2550 tasks per day, so I think these percentages are low. I run 90% Zika WUs, 8% MIP and 2% other tasks.

So, having checked everything that people have suggested, I think there is something at the account level preventing me from getting extra work. I find it hard to believe that
14 computers over 2 OS's will suddenly all lose their buffers at the same time, without me tweaking any settings at all.

Additionally, the two linux based 80 thread machines only get 64 tasks generally. Even though at startup, boinc does see all 80 threads and boinc stats also reports both machines at 80 threads.

Am I missing anything? I really want to see if an admin can check my account, and if they can, how do I get in touch with them?

Lavaflow,
They all have several gigs of space free. Even if that would be the issue, it would not hit all 14 machines at the same time.

Hence I am asking what is the best way to ask an admin to look into my account.

We're very much interested what the answer is. To expand on how BOINC looks at space
10GB allowed,the disk is 100GB large and I'm allowing 60% use of it, means only 6GB is really permitted.

Anyway, the message log should be telling if you hit update what if any work if requested and hint at why it is not sending more work. Set some extra log flags such as CPU scheduler <cpu_sched>1</cpu_sched> and debug <sched_op_debug>1</sched_op_debug> and <sched_ops>1</sched_ops>

You'd get log entries such as below down to the second

6/12/2019 8:18:00 AM Sending scheduler request: To fetch work.
6/12/2019 8:18:00 AM Requesting new tasks for CPU
6/12/2019 8:18:00 AM [sched_op] CPU work request: 72681.19 seconds; 0.00 devices
6/12/2019 8:18:03 AM Scheduler request completed: got 0 new tasks
6/12/2019 8:18:03 AM [sched_op] Server version 701
6/12/2019 8:18:03 AM No tasks sent
6/12/2019 8:18:03 AM No tasks are available for Microbiome Immunity Project
6/12/2019 8:18:03 AM No tasks are available for OpenZika
6/12/2019 8:18:03 AM No tasks are available for Help Stop TB
6/12/2019 8:18:03 AM No tasks are available for the applications you have selected.
6/12/2019 8:18:03 AM Project requested delay of 121 seconds

The 'no work' is because each has hit the max units allowed in my device profile. 8, 6, 13

Another thought, if the custom section in device profiles is set to allow 10 units for each project and you choose to crunch on only 1, then your cache will never be greater than 10 at that time. If all devices run off the same profile, that could be the common denominator.

Excellent. Possibly the many reruns of the period end process yesterday by techs unlodged something, but sense it doesn't make.