Message boards : LHCb Application : VirtualBox 5.1.x vs. 5.2.x
Message board moderation

To post messages, you must log in.

AuthorMessage
computezrmle
Volunteer moderator
Volunteer developer
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 15 Jun 08
Posts: 2534
Credit: 253,851,031
RAC: 37,958
Message 35182 - Posted: 7 May 2018, 7:25:10 UTC

For a couple of weeks I notice that lots of my LHCb VMs suffer from very short runtimes.
Although nearly all of them get rewarded, it is an unsatisfying situation as it seems that those WUs don't really deliver scientific output.

Thus I tried to compare the efficiency of my hosts with that of other volunteers' hosts (top 40 LHCb contributors; data available to the public).
The results were not 100% decisive but gave me a strong hint that VirtualBox 5.2.x may be a factor.
I therefore downgraded VirtualBox to version 5.1.36 (the most recent 5.1.x).

All WUs I ran with VirtualBox 5.1.36 over the last weekend on different hosts had normal runtimes and the size of the result uploads is also back to normal.

@other volunteers
Did anybody made the same experience?
ID: 35182 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 2243
Credit: 173,902,375
RAC: 2,013
Message 35185 - Posted: 7 May 2018, 7:48:18 UTC

Had one PC (AMD-FX) that crashed with Windows Fall-Creator and Virtualbox 5.1.34.
On Boinc-Website is Virtualbox 5.1.26 with Boinc 7.8.3 the default.
This is working now for me after downgrading.
ID: 35185 · Report as offensive     Reply Quote
Jim1348

Send message
Joined: 15 Nov 14
Posts: 602
Credit: 24,371,321
RAC: 0
Message 35186 - Posted: 7 May 2018, 12:14:57 UTC - in response to Message 35182.  
Last modified: 7 May 2018, 12:23:51 UTC

I just started up LHCb again on an Ubuntu 16.04 machine with VBox 5.2.10. Of the first 12 returned, 5 were short (including one error).
https://lhcathome.cern.ch/lhcathome/results.php?hostid=10544450&offset=0&show_names=0&state=0&appid=12

Also note that on the short ones, the CPU usage was low, less than about 25 percent.

But I currently have seven in process that have gone over 2 hours, so they seem OK. If the shorts remain too high, I will go back to VBox 5.1.x.
Thanks for catching it.
ID: 35186 · Report as offensive     Reply Quote
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 847
Credit: 691,699,382
RAC: 114,148
Message 35219 - Posted: 10 May 2018, 9:34:31 UTC

I would agree that 5.2.x isn't as reliable but I have had good results from LHCb with 5.2.
ID: 35219 · Report as offensive     Reply Quote

Message boards : LHCb Application : VirtualBox 5.1.x vs. 5.2.x


©2024 CERN