It has been running for about 1h now. CPU usage is close to 100% all the time( except a few dips for a few seconds every now and then)

It has been running for about 1h now. CPU usage is close to 100% all the time( except a few dips for a few seconds every now and then)

After 12 hours the Task Manager is still showing 0 to 2% total with nothing else going on. It would appear something is amiss but at this point, since I am over half done the task, I will wait and see what shakes out when it completes. Perhaps there will be an error message to explain further what is happening. Thanks for letting me know of your CPU usage.

I don't have the required download capacity either. Best of luck Ivan.

---

Pluto will always be a planet to me.

Seti Ambassador
Not to late to order an Anni Shirt

I'm probably an average user although I'm paticipating in WOW.
approximatly 15K for the two week event if I'm dedicated.

Does the machine need to run 24/7 while we're contributing for CMS?

I can't get any work from cms

---

[Edit] Oh, to see the VBox display you'll also need the appropriate Extension Pack for your version of VirtualBox -- download here. [/Edit]

Extension pack for 4.3.30, BOINC installed 4.3.12. And extension pack refused to install on this version. Worth to note.

Another note: with BOINC's default of 10GB and some other projects like SETI & Einstein installed, your one refuse to send work:

17-Aug-15 11:18:06 | CMS-dev | Message from server: CMS Simulation needs 16.38MB more disk space. You currently have 9520.37 MB available and it needs 9536.74 MB.


It's ok for testing to change BOINC defaults but could be problematic in wider run.

And third note: upon task start Windows Firewall blocked: C:\program files\oracle\virtualbox\vboxheadless.exe executable from net access. Required manual approval. Again, not good for usual run (but this more part of BOINC/VM interaction, perhaps).

Ok, I'm in. cmsRun occupies top of top :), but not constantly. CPU usage very unstable though BOINC reserved one of 4 cores for it.

And it sems pushing "Graphics" button on the left leads to web page with logs for my host. Nice approach!

And another observation: VBoxHeadless runs @normal priority. That is, on higher priority than default for GPU SETI tasks. Also, it doesn't use affinity so can land on any core, including one with SETI GPU task. I'm afraid it could negatively affect GPU performance time to time.

==== PYTHIA WILL USE LHAPDF ====
*************************************
* LHAPDF Version 5.9.1 *
* Configured for the following: *
* All PDFs *
* LOW MEMORY option *
* Maximum 1 concurrent set(s) *
*************************************

Interesting, how to configure it with high memory and will it speedup things? Host has 8GB RAM.

Seems one data chunk completed:

TimeReport> Time report complete in 7846.97 seconds
Time Summary:
- Min event: 0.00409007
- Max event: 103.886
- Avg event: 7.84697
- Total job: 7846.97
CPU Summary:
- Min event: 0.004
- Max event: 39.969
- Avg event: 7.15856
- Total job: 7203.76
- Total event: 7158.56

On next will try to suspend/restart.

EDIT: and here inefficiency starts: BOINC's task continues to run. So core still reserved. But VM runs non-CPU intensive script to rearrange next data chunk processing/data renewal. These CPU cycles are lost.
So, one needs to run BOINC in overcommitted mode to fully utilize CPU in this case.

Question: I exited from top (pressing q) so how to re-run it w/o VM restart? Discovered console on Alt-F6 but don't know root password. Something interesting on Alt-F10 also.

Suspending task just paused VM, restarting BOINC resulted in taking VM's snapshot and then resuming from it. So, I returned into console w/o top again :/.

And now, with second job, job's log file doesn't get updates. So - blind run now. Maybe worth to open VM w/o BOINC and reboot it?

Recovery after VM shutdown still goes...

EDIT: and here inefficiency starts: BOINC's task continues to run. So core still reserved. But VM runs non-CPU intensive script to rearrange next data chunk processing/data renewal. These CPU cycles are lost.

Yes, I seem to remember having a conversation like that with Ben Segal in London, in 2010 (when the early development work on the CERN VM extensions was first reported at a BOINC workshop). I suggest you take it up with Laurence on the CMS-dev boards - Ivan isn't really responsible for that part of the process, and it isn't a SETI issue either.

Question: I exited from top (pressing q) so how to re-run it w/o VM restart? Discovered console on Alt-F6 but don't know root password. Something interesting on Alt-F10 also.

Suspending task just paused VM, restarting BOINC resulted in taking VM's snapshot and then resuming from it. So, I returned into console w/o top again :/.

And now, with second job, job's log file doesn't get updates. So - blind run now. Maybe worth to open VM w/o BOINC and reboot it?

Yeah, I'm not sure about that, except to say "don't press Q!". Also, obviously, I'm not going to give out the password for logon indiscriminately.

Feel free to play around with the VM as you see fit, we're not in production yet and can afford to lose results to experimentation.

---

and it isn't a SETI issue either.

LoL, it's SETI issue as long as SETI don't receive CPU cycles ;D :D.

Well, Ivan asked to test so I will post here little more. This VM thing under BOINC is new to me so yep, expect some already known/silly appears :)
Interesting approach though.

Feel free to play around with the VM as you see fit, we're not in production yet and can afford to lose results to experimentation.

Yeah, new toy :D :D

P.S. Seems it recovered from shutdown (maybe such shutdown will emulate main host shutdown, where BOINC can't gracefully take snapshot before OFF).
csmRun using CPU again. Will see if it puts smth in web log...

Raistmer, I see you're attached to CMS-dev with a computer running BOINC v7.4.42

Please could you - and anybody else following Ivan's invitation in this thread - heed the warnings and upgrade any machines attached to CMS-dev to at least BOINC v7.6.2 (ideally v7.6.6 with yesterday's 160815 hotfix). Otherwise, you may fall foul of the >4 GB file deletion bug, and trash work at SETI and other projects too.

Remember that CMS-dev is still very much an experimental project, and all the usual advice about Beta testing applies.

Raistmer, I see you're attached to CMS-dev with a computer running BOINC v7.4.42

Please could you - and anybody else following Ivan's invitation in this thread - heed the warnings and upgrade any machines attached to CMS-dev to at least BOINC v7.6.2 (ideally v7.6.6 with yesterday's 160815 hotfix). Otherwise, you may fall foul of the >4 GB file deletion bug, and trash work at SETI and other projects too.

Remember that CMS-dev is still very much an experimental project, and all the usual advice about Beta testing applies.

OK, I'll dive little deeper.

Hm... 7.6.6 installer killed boinc.exe and all apps... but VBoxHeadless one. It has 3 instances currently, with only 1 using CPU.
Looking via RDP Cern's VM is alive. What will be if I just start BOINC manager now?... will see.

EDIT: It marked task as: Postponed: VM environment needed to be cleaned up.

So just will do host reboot.

It has 3 instances currently, with only 1 using CPU.

I have the same situation.

I asked the invitation code, got it, but the BOINC manager 7.4.42 on Windows 10 would not let me create an account. So I went to the home page, saw "create an account" and created it. Went back to the BOINC manager and added it to my BOINC projects. I got a task and is running now.Virtual Box 4.3.30.
Tullio

---

LoL, I like it :D

(PYPTIS:) Sorry, I got a heavy companion quark here. Not handled yet, giving up!