Earlier in the week, we experienced a drought of work units across all projects, despite none of our back-end systems indicating that anything was wrong (and we had plenty of MCM1, SCC1, ARP WUs in the pipeline). On Tuesday, some MCM1 and SCC1 WUs were being distributed by our system, albeit at a much slower rate than usual.

We have taken measures to send out work in spite of the distribution issues. For MCM1 work units, there were 2.1 million MCM1 work units ready to be distributed, but only 7 work units were actually loaded and waiting to be sent. To increase this number, our bottleneck is in a “create-work” process which simultaneously creates new workunits and processes the existing ones so a healthy amount are is in circulation at any time. The create-work process is currently loading 100 work units into BOINC every 17-20 seconds, which we further reduced to 13 seconds using parallel threads.

We are also looking to load more SCC1 work units as they appear to be exceedingly rare currently. The distribution rate of ARP1 work units seem to be inconsistent and rare, so we are looking into the possible cause. The long period of pending validation for ARP1 work units was due to the ARP1 validator, which was taking longer to validate the high number of results returned. Using parallel threads, we reduced the time the validator takes and, accordingly, the pending validation time. OPN1 and OPNG are still not available, as the OPN team is conducting new research before sending us more (as stated in our May 2023 workunit update).

If you have any comments or questions, please leave them in this thread for us to answer. Thank you for your patience.

WCG team