Great update on ARP by Savas (THANK YOU !!!). https://www.worldcommunitygrid.org/forums/wcg/viewpostinthread?post=699984
I love all the details and the fact that I now know they are still working on the issue.

I'm now seeing a new line in my event log that I find interesting.
Tue 19 Nov 07:54:26 2024 | World Community Grid | Tasks won't finish in time: BOINC runs 51.2% of the time; computation is enabled 78.0% of that
This is false. I'm finishing all WUs it gives me in plenty of time. the numbers are going up, so I think it is an average of data, and I took a week off, so it is slowly increasing. I think it limits what WUs I can download, which is frustrating.

Mike - thanks for the response...
Agreed! Interestingly, the two-day "completed" total usually runs very close to the number of units that can be shown to have moved [via generations.txt changes] over the same two days, but sometimes it is off by more than can be accounted for by the scripts running at slightly different times :-)
Thank you -- I thought it was on sending out, but wasn't sure!
Fair enough -- that's why I place my trust in the same file :-). The only use I can think of for the completed.txt file is that it does give some insight into possible retry counts (though that's not possible when uploads and downloads can take days...)

Hopefully, once most/all of the cells have their current workunits in the right categories it might even be possible to have a guess at what's missing from state.txt (and why). However, it will only be a guess :-)

Once again, thanks -- I'd been half-hoping you might have been privy to some [off-board] communication that clarified further, but...

Cheers - Al.

Hi Unixchick,

No, despite now having a few 'slots' open and available for ARP WUs, they're currently awaiting new work. Don't worry - MCM is 'filling the gap'

Edit
I've just received a repair job ARP1_0009078_143_2

I am now totally out of ARP. No downloads or uploads or cache.

Mike

TL/DR -- Sgt. Joe has it right, both about [extended] suspension of work and the improved upload (and download) speed!

According to the posts by savas (which I would encourage more people to read!), throughput at the load balancer was a lot lower than expected and various changes seem to have caused a significant improvement there. They [still] intend to try to sort out a release schedule for ARP1 tasks that is less likely to cause issues (even at the improved network throughput currently being observed!) -- I love the references to titration in the messages [dosage strategies?] :-)

As for the current pause, there is a certain amount of time required for housekeeping of returned results and preparation work for the next tranche of work anyway, resulting in a delay before new WUs could be generated; it remains to be seen how soon after WU generation has started they will be prepared to restart the feed.

By the way, I believe that the work done processing returned results and preparing new work can take place whilst there are still existing WUs waiting to be served up, in which case there ought to be a balance point at which we wouldn't see pauses in work availability; however, for that to work well they might need to [partially] hamstring some users who run buffers that pick up several days-worth of work, and that might be dangerous ground[*1]! :-)

On a slightly related note, I wonder if they may also consider a slight alteration in the work issue pattern for OPNG the next time it shows up; it isn't clear whether OPNG on its own would still cause the sort of upload and download issues we've seen in the past (given the increased active bandwidth) but I hope they'll keep an eye out if/when OPNG resurfaces...

Cheers - Al.

*1 - Declaration of bias on my part! -- for ARP1 I'd rather return stuff quickly (i.e. within a day if possible) and run the risk of running out of work if there's a system problem. I'm here for the science, not the RAC or points :-)