Yes. The Macs are all in my basement, near each other. Normally (when WCG works), they just crunch away 24x7 without any issues. Power goes out?...they start right back up without human intervention. If I notice 1 machine hasn't submitted results in a few days (this doesn't happen on the Macs...only on my Wintels when Windows decides to have a nag screen after a Windows Update), I just power it off and on and it's back up. If for some bizarre reason the Mac continues not to report results, I will connect my kb and monitor.

The pain is circumstances like right now...I don't want to connect a kb and monitor to each of 15 Macs...click Retry...wait 1-4 mins...likely have to click Retry again...wait 1-4 mins...then when/if it works, unplug and go to the next Mac. It's just a total time waster.

So it appears in this case I can just let them sit and this issue will (hopefully) fix itself within the next few days.

Having 16 real cores, I can be downloading quite a few ARP tasks at a time. It would be nice if BOINC could concentrate on the downloads for one task at a time, that could be helpful. I often seem to have four or more tasks downloading with only 6 files left.
(This has been put in as a feature request in the past to git-hub. Not sure why it has been rejected.)

If anyone is listening, it's 10am ET and the results from yesterday for all 15 machines are still about 1/3 of what they normally are...so this problem is not fixed yet. :(

I know this has been mentioned before, but is there any technical reason why the files cannot be combined into one file for download?

If every work unit is ONE file, then:

1. Once a download for a work unit begins because a connection was made, then it SHOULD complete (no more work units that are missing files).
2. About 10x fewer connection requests to the servers allows more work to move freely.
3. More completed tasks = more valids = fewer resends = less traffic.

Is there a technical limitation of the project that prevents this from being a possibility?

I would be THRILLED to download one file, even if it is large and takes a while, versus the fragmented process. Some other projects out there require the download of massive files, and few participants complain about file size.

My guess the challenges are:

1)The server would need to combine/zip/rar all your tasks into a bundle...requiring more/a lot of CPU time for thousands of users.

2)Possibly file size and hence where to store "the bundle" which is now essentially 2x the storage needed for WCG.

3)WCG projects' bundles could have massive file sizes


An answer to #1 is that simply the Server gives every client a "bundle" of X WUs. But what is X? I don't think WCG wants to keep Small, Medium, and Large bundles all over their storage and constantly try to predict when the bundles are starting to go dry and then some users don't get bundles for a few days.

Ideally as I've mentioned before, Krembil should have just re-written this relatively simply client/server app in modern code and plop it on a modern Cloud environment. Heck, the darn website takes 10 seconds to load ANY non-stats page while the stats pages take far longer! I bet you this website gets about 1000 visits an hour on any given day which is tiny.

There is no reason stats pages should take 20+ seconds to load in 2004 let alone 2024. The stats areas of this site should either be 1)re-written to show stats as of 24 hours ago and hence the stats are very simply db lookups into static summary Table entries or 2)the databases should be put on far faster CPUs, storage, and/or RAM as well as possibly a better db platform and/or better Table and/or db design. I sense the db is on IBM DB2 (which has a world of drawbacks compared to other dbs), was simply migrated from one OS to a new OS, AND is on low-performance hardware. I'm all ears to hear more details about how the technology behind WCG if someone has it.