LHC@home

Ivan, thanks a lot for the information :-)

the queue has run dry :-(

the queue has run dry :-(

Oops! Apologies. I edited the workflow script to halve the number of jobs per submission, but with my deteriorating vision[*] I deleted one too many digits and ended up with only a tenth of the jobs I'd intended.
[*] Look up dry age-related macular degeneration. See also: cataracts, hayfever.

---

I'm about to close off the current workflow to allow a WMAgent upgrade. You might want to set No New Tasks until it's over.

---

Thanks for the update Ivan

(and -dev)

I'm about to close off the current workflow to allow a WMAgent upgrade. You might want to set No New Tasks until it's over.

It's done. Barring problems, jobs should be coming down the turnpike Real Soon Now.

---

the queue has run dry :-(

the queue has run dry :-(

It's strange, WMStats insists there are ~1200 jobs running, Grafana says there are none.

---

An upgrade went wrong. Seems to be working again now.

---

Seems to be working again now.

Right.
Thanks.

Oops, holiday here, plus I've been dealing with a neighbour whose daughter comitted suicide this morning...
New jobs on their way.

---

There's a database problem at CERN which means I can't submit new jobs. We will run out of jobs shortly if it's not fixed.

---

New jobs on their way, but it will take an hour or so to get them into the queues.

---

Any reason why my computers set to get any type of work (they usually get CMS, I assume that's your top priority) were told there were no tasks, even though I can see Atlas and Theory available? The ones set to not get CMS continued to get Atlas and Theory ok.

Any reason why my computers set to get any type of work (they usually get CMS, I assume that's your top priority) were told there were no tasks, even though I can see Atlas and Theory available? The ones set to not get CMS continued to get Atlas and Theory ok.

It could be that your quota has run down -- I see one of your machines had a lot of aborted tasks:
https://lhcathome.cern.ch/lhcathome/results.php?hostid=10825779&offset=0&show_names=0&state=0&appid=11
Each time you return an error task, your daily quota is reduced, and it will eventually reach one per day so much of the time it will tell you that there are no tasks (although it should say you've reached quota in the event log). Another possibility is that there is a problem with the VirtualBox setup so BOINC won't send tasks that require a VM, but your machines all say virtualisation is enabled.

---

Any reason why my computers set to get any type of work (they usually get CMS, I assume that's your top priority) were told there were no tasks, even though I can see Atlas and Theory available? The ones set to not get CMS continued to get Atlas and Theory ok.

It could be that your quota has run down -- I see one of your machines had a lot of aborted tasks:
https://lhcathome.cern.ch/lhcathome/results.php?hostid=10825779&offset=0&show_names=0&state=0&appid=11
Each time you return an error task, your daily quota is reduced, and it will eventually reach one per day so much of the time it will tell you that there are no tasks (although it should say you've reached quota in the event log). Another possibility is that there is a problem with the VirtualBox setup so BOINC won't send tasks that require a VM, but your machines all say virtualisation is enabled.

The machines in question had only been running CMS, and had not produced errors of any significant quantity, and were never refused CMS. They were set to get any type of task, but always receive CMS (I assume that's the most important one to you). When I saw CMS run out, I tested to see if they'd get something else automatically. I forced them to get new work, and they were told there was none. Yet the computers I told not to get CMS (different venue, to stop me running out of internet bandwidth) were happily getting Theory.

I've seen this before, in that the server just seems to send all of one sub-project, to me it seems to be a server side thing.

I personally came to the conclusion that the scientist or administators have the balance of work the way they want it and I should not try to make an even split between the sub-projects.

I've seen this before, in that the server just seems to send all of one sub-project, to me it seems to be a server side thing.

I personally came to the conclusion that the scientist or administrators have the balance of work the way they want it and I should not try to make an even split between the sub-projects.

It may well be a lot of people choose one subproject for RAM/bandwidth/personal reasons, and that happens to even out the way the scientists need it. If everyone selected "anything", only CMS would get done. But most folk don't have the RAM or bandwidth to run those. I just run them on 2 of my 8 machines, and that's my uplink saturated. I set the rest to run "anything but CMS" and they get theory with the occasional Atlas.

the queue has run dry :-(

the queue has run dry :-(

it has filled up now :-)