LHC@home

Check out https://lhcathome.cern.ch/lhcathome/results.php?hostid=10520845 Notice of 1,741 tasks issued it has returned 1,651 errors. Fubar host of the year? Not really. Drill down to host's Sixtrack tasks and see 1,254 errored tasks most of which were returned < 60 seconds after they were sent and have status "Not started by deadline - canceled. What am I seeing here... tasks with a deadline of < 60 seconds? Task details say aborted by user but somehow I doubt that. Sounds more like major server error.

Now drill down to host's ATLAS tasks and notice 119 ATLAS tasks all errors! That's 119 X 350MB downloads pissed down the drain. And if you take time to look you can find dozens more such hosts. Does anybody still not understand why ATLAS downloads are only ~200KBps?

Are there any admins who still can't see that the server side option to restrict dl's to hosts returning steady errors needs to be checked?

One possible scenario I see why this could have happened is if the computer's clock is set wrongly and shows some date in the future. I checked a couple of new tasks on that host and I see in the task display that the deadline is not until 23rd of August. There is a function to automatically abort tasks that were not started by deadline, whether that is built in Boinc or just in BoicTasks (which I use instead of Boinc Manager), I don't know. But that could explain what happened with this host.

Anyway, this explanation does not remove the need to restrict the misbehaving hosts from downloading huge amounts of new tasks after thrashing a buttload of them first.

---

That would make sense. So fubar host not fubar server. Maybe they're struggling with the concept of overclocking.
Well, dude's gittin somethin' from bronco and no, it's not a Dell.
<edit>
Another misconfigured Gridcoin host. A pattern seems to be emerging.
</edit>