Development for LHC@home

But our computers request a certain number of tasks.
They were sent out at the same time, because the are essentially zero length.

They would still have to have removed the limit at the server end restricting tasks to one. If that was removed I assume it would resort to the cache settings ?

Correction, the clients request a certain number of seconds, and the server works out, how many tasks to send.

Any ideas what is causing this behaviour?

As far as I can tell, we're still sending out jobs. Which implies we're sending out tasks, unless there are a lot of tasks still within their 24+ hour limit. I'll look up the recipe and send more tasks to CMS-dev.

---

Does this also happen with CMS-simulation tasks at vLHC?

Does this also happen with CMS-simulation tasks at vLHC?

We don't know, because VirtualLHC@Home doesn't have CMS-tasks during the last 2-3 days.
If it does, it should also affect vLHC-tasks and that doesn't happen.
So it's a CMS-dev server issue.

As far as I can tell, we're still sending out jobs. Which implies we're sending out tasks, unless there are a lot of tasks still within their 24+ hour limit. I'll look up the recipe and send more tasks to CMS-dev.

The very few tasks sent out are resend tasks, because aborted, compute error or timed-out after the 7 days deadline.
The 24 hour limit has changed to 36 hours. The graceful shutdown after a run finished after 24+ hours doesn't work.

I just got told I needed more space too, event log shows...

05/02/2016 11:16:26 | CMS-dev | Message from server: CMS Simulation needs 5788.37MB more disk space. You currently have 3748.37 MB available and it needs 9536.74 MB.
05/02/2016 11:16:28 | CMS-dev | Started download of CMS_2016_01_28.xml
05/02/2016 11:16:28 | CMS-dev | [file_xfer] URL: http://boincai05.cern.ch/CMS-dev/download/CMS_2016_01_28.xml
05/02/2016 11:16:28 | | [http_xfer] [ID#2091] HTTP: wrote 514 bytes
05/02/2016 11:16:29 | CMS-dev | [file_xfer] http op done; retval 0 (Success)
05/02/2016 11:16:29 | CMS-dev | [file_xfer] file transfer status 0 (Success)
05/02/2016 11:16:29 | CMS-dev | Finished download of CMS_2016_01_28.xml
05/02/2016 11:16:29 | CMS-dev | [file_xfer] Throughput 3807 bytes/sec

Boinc itself is correctly reporting there is over 30Gb of space to use

I got two (vlhc-tasks) yesterday.
If i request work here, i get disk low warning "needs 9000Mb" or so.
If i request it at vLHC, i do not get that message.
Both have 0 tasks to send, but only cms is giving that message.
(disk space is not low)

I still haven't received a task to run on my laptop but the server shows I have 16 so these new tasks will quickly disappear again :-(

Edit: Every time I click Update I get more phantoms on my list but nothing is received.

Any ideas what is causing this behaviour?

As far as I can tell, we're still sending out jobs. Which implies we're sending out tasks, unless there are a lot of tasks still within their 24+ hour limit. I'll look up the recipe and send more tasks to CMS-dev.

Well, create_work + make_work is still chugging away on boincai05, but tasks-ready-to-send is now showing 5001. :-) I'm not convinced it will make a difference, tho'-but...

---

I tried turning some logging options on to see if anything else useful might be there...

05/02/2016 12:13:54 | CMS-dev | [work_fetch] set_request() for CPU: ninst 4 nused_total 0.00 nidle_now 2.00 fetch share 1.00 req_inst 4.00 req_secs 1406952.06
05/02/2016 12:13:54 | CMS-dev | [sched_op] Starting scheduler request
05/02/2016 12:13:54 | CMS-dev | [work_fetch] request: CPU (1406952.06 sec, 4.00 inst)
05/02/2016 12:13:55 | CMS-dev | Sending scheduler request: Requested by user.
05/02/2016 12:13:55 | CMS-dev | Requesting new tasks for CPU
05/02/2016 12:13:55 | CMS-dev | [sched_op] CPU work request: 1406952.06 seconds; 4.00 devices
05/02/2016 12:13:55 | | [http_xfer] [ID#0] HTTP: wrote 1180 bytes
05/02/2016 12:13:55 | | [http_xfer] [ID#0] HTTP: wrote 4198 bytes
05/02/2016 12:13:55 | | [http_xfer] [ID#1] HTTP: wrote 1298 bytes
05/02/2016 12:13:55 | | [http_xfer] [ID#1] HTTP: wrote 5183 bytes
05/02/2016 12:13:56 | CMS-dev | Scheduler request completed: got 0 new tasks
05/02/2016 12:13:56 | CMS-dev | [sched_op] Server version 707
05/02/2016 12:13:56 | CMS-dev | No tasks sent
05/02/2016 12:13:56 | CMS-dev | Message from server: CMS Simulation needs 5787.37MB more disk space. You currently have 3749.37 MB available and it needs 9536.74 MB.

No tasks received but list on the server grows longer.

Tasks in progress has gone up fro 906 to 1136 in about 15 min.
But "Tasks ready to send" remains at 5001.????

Detached and re-attached the CMS-dev, but fruitless.
As I hoped the old ghosts are gone (abandoned), but have 32 new ghosts and no single one on my computer.

Because of the irrelevant disk space messages, I tried increasing what normally makes no sense (31GB free for BOINC) with the next messages:

05 Feb 13:34:46 max disk usage: 40.00GB
05 Feb 13:35:58 Message from server: CMS Simulation needs 5218.37MB more disk space. You currently have 4318.37 MB available and it needs 9536.74 MB.
05 Feb 13:42:03 max disk usage: 100.00GB
05 Feb 13:42:42 Message from server: CMS Simulation needs 2406.63MB more disk space. You currently have 7130.11 MB available and it needs 9536.74 MB.
05 Feb 13:44:14 max disk usage: 200.00GB
05 Feb 13:45:00 Message from server: CMS Simulation needs 4923.73MB more disk space. You currently have 4613.02 MB available and it needs 9536.74 MB.

After the new attach to the project and the first MB-needs message, 3 tasks were added to my list, after the second request 9 and after the 3rd request 20 tasks were added.

Edit: 05 Feb 14:37:58 [disk_usage] allowed 40960.00MB used 9476.97MB
19830 CMS-dev 05 Feb 14:37:58 [disk_usage] usage 1377.75MB share 1553.12MB
19861 CMS-dev 05 Feb 14:37:58 Sending scheduler request: Requested by user.
19862 CMS-dev 05 Feb 14:37:58 Requesting new tasks for CPU
19863 CMS-dev 05 Feb 14:38:00 Scheduler request completed: got 0 new tasks
19864 CMS-dev 05 Feb 14:38:00 No tasks sent
19865 CMS-dev 05 Feb 14:38:00 Message from server: CMS Simulation needs 6663.95MB more disk space. You currently have 2872.80 MB available and it needs 9536.74 MB.

Yes, I'm getting the same message.

---

Yes, I'm getting the same message.

If I wouldn't be so innocent, I would think CMS-dev has found a manner to push all of us over to VirtualLHC@home to crunch CMS-tasks there :P

I tried to get a CMS task at vLHC and couldn't, that was after more jobs had been released and appearing in the queue (and being taken by others) !

There is CMS work now at vLHC, and it is working properly.

That are not mine hosts, I've 'only' 16 virtual tasks and 1 real task in progress.

I know, that's why you are 3rd and I'm now 4th.

Since new tasks were loaded, we have a new competition and maybe a new leader: Host with 381 tasks in progress.

I think i know, why the Jobs graph is so screwed up.
It is caused by the WMAgent backfill jobs, that are ALL failing.

Any news on fixing the server?