Message boards :
ATLAS application :
ATLAS vbox version 2.00
Message board moderation
Author | Message |
---|---|
Send message Joined: 13 May 14 Posts: 387 Credit: 15,314,184 RAC: 0 |
After two and a half years of version 1.01 we are happy to announce a new version of ATLAS vbox! This version updates the image from SLC6 to CentOS 7, and the bootstrapping scripts have been improved to give faster startup times and better logging. The image is a bit larger than the old one, 2.7GB (1.1GB compressed download) and therefore the disk space limits have been increased to 10GB per task. This new version has been extensively tested on LHC-dev, many thanks to all the volunteers who helped there! However since this is such a major update we have made it a beta version, so you will only get it if you have selected "Run test applications" in your preferences. Please try it out and give your feedback! EDIT: the server was not distributing any WU with the new version so I removed the beta flag. Now everyone should get it. |
Send message Joined: 14 Jan 10 Posts: 1413 Credit: 9,435,474 RAC: 8,120 |
I suppose this new versions (native and vbox) will be sent, when all older unsent WU's have been distributed - 3264 tasks at the moment. |
Send message Joined: 13 May 14 Posts: 387 Credit: 15,314,184 RAC: 0 |
No, it should be effective immediately. The decision on which version to use is taken when the client asks the server for new tasks. However I see no WU have been sent which use the new versions yet... BOINC's choice of app version has always been a mystery to me and usually I deprecate previous versions to force new versions to be used. But first I will take the new versions out of beta to see if it helps. |
Send message Joined: 27 Sep 08 Posts: 831 Credit: 688,871,932 RAC: 126,472 |
I got one, the task size is still kind of messed up like the 1.01. I have a 8day task at 43GFLOPS and the new one that 99.999% in 5hr or less, they can be both 43GFLOPS on the same computer. |
Send message Joined: 15 Jun 08 Posts: 2521 Credit: 252,581,430 RAC: 137,194 |
On my computers all ATLAS tasks fail since x86_64-centos7.img is used as singularity image. A few tasks running v2.72 native are running fine with image x86_64-slc6.img. |
Send message Joined: 13 May 14 Posts: 387 Credit: 15,314,184 RAC: 0 |
On my computers all ATLAS tasks fail since x86_64-centos7.img is used as singularity image. Not sure I understand - 2.72 uses the centos7 image and all previous versions used the slc6 one. Could you give some examples? You have a lot of hosts and it's a lot of clicking to find the ones which run ATLAS :) And maybe this would be better discussed on the thread for the new native version |
Send message Joined: 15 Jun 08 Posts: 2521 Credit: 252,581,430 RAC: 137,194 |
And maybe this would be better discussed on the thread for the new native version You are right. Sorry. Meanwhile the problem seems to be solved: https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=5161&postid=40110 |
Send message Joined: 27 Sep 08 Posts: 831 Credit: 688,871,932 RAC: 126,472 |
Do you have an idea when they might finish? So far they reached 100% in 1-2hrs but the FLOPS estimate must be way wrong since they are still going after a day or so. Do you have a way to edit the FLOPS estimate? |
Send message Joined: 7 May 08 Posts: 205 Credit: 1,564,905 RAC: 1,765 |
Do you have an idea when they might finish? Same here. Wu arrived quickly to 99%, then slowed down and now, after 22hs is at 100%, but still crunching. Is it normal? |
Send message Joined: 2 May 07 Posts: 2230 Credit: 173,863,695 RAC: 17,252 |
Have 1.01 and 2.00 running. Up to 250k seconds with 5 or 6 CPU's. Up to 20 Minutes per collision (200 Collisions). |
Send message Joined: 15 Jun 08 Posts: 2521 Credit: 252,581,430 RAC: 137,194 |
Toby Broom wrote: Do you have an idea when they might finish? boboviz wrote: Same here. This is the usual behaviour whenever a new app version is introduced. BOINC uses 2 main factors to calculate estimated runtimes (as well as credits): - estimated GFLOPS for a task - peak GFLOPs of a computer Although it is a BOINC recommendation to estimate the task's GFLOPS as accurate as possible before the server sends it to a client, ATLAS always uses a fixed value. Based on that value together with the returned runtime/CPU-time the server calculates an average processing rate for each host and stores it in the DB. The server returns that value to the client when fresh work is sent. A new app version usually starts with the benchmarked processing rates and needs some days or even weeks to get enough returned results. As an example you may compare the average processing rates for ATLAS 1.01/2.00 on this host: https://lhcathome.cern.ch/lhcathome/host_app_versions.php?hostid=10567798 Toby Broom wrote: Do you have a way to edit the FLOPS estimate? Wouldn't make much sense: - The adaptive algorithm solves the issue - ATLAS events have variable runtimes even inside a task It might be possible to estimate an average for a given ATLAS batch but there are usually a couple of batches in the queue and tasks are sent out in random order. |
Send message Joined: 27 Sep 08 Posts: 831 Credit: 688,871,932 RAC: 126,472 |
- ATLAS events have variable runtimes even inside a task So does SixiTrack but this doesn't not leave tasks running for 6days at 99.999, so there is a possibility to configure the tasks better. |
Send message Joined: 7 May 08 Posts: 205 Credit: 1,564,905 RAC: 1,765 |
Wu arrived quickly to 99%, then slowed down and now, after 22hs is at 100%, but still crunching. Finished!! (after 34hs) |
Send message Joined: 25 Sep 17 Posts: 99 Credit: 3,425,566 RAC: 0 |
Are we supposed to get any info in the second terminal within the running virtual machine? Alt + F2? I see TOP related info in Alt + F3 but just login screens on the first and second terminal screens while I have activity on all eight assigned processors within the VM (athena.py) Nevermind, I started to see info. It just took a while. |
Send message Joined: 24 Oct 04 Posts: 1169 Credit: 54,428,146 RAC: 61,850 |
Wu arrived quickly to 99%, then slowed down and now, after 22hs is at 100%, but still crunching. And 3,523.68 Credits |
Send message Joined: 15 Jun 08 Posts: 2521 Credit: 252,581,430 RAC: 137,194 |
Wu arrived quickly to 99%, then slowed down and now, after 22hs is at 100%, but still crunching. The task finished and got credits - that's BOINC's point of view. But you may be aware that from the scientific point of view the task didn't succeed: 2019-10-12 05:27:54 (12056): Guest Log: No HITS file was produced Your logfile shows that the task started/paused several times and couldn't successfully write it's snapshot: 2019-10-10 07:23:49 (13064): Stopping VM. 2019-10-10 07:23:58 (13064): Error in stop VM for VM: -108 Command: VBoxManage -q controlvm "boinc_0c99027d1c3c10c0" savestate Output: 0%...10%...20%...30%...40%... 2019-10-10 07:23:58 (13064): VM did not stop when requested. 2019-10-10 07:23:58 (13064): VM was successfully terminated. . . . 2019-10-10 18:46:27 (6124): Stopping VM. 2019-10-10 18:46:29 (6124): Error in stop VM for VM: -108 Command: VBoxManage -q controlvm "boinc_0c99027d1c3c10c0" savestate Output: 0%...10%...20%... 2019-10-10 18:46:29 (6124): VM did not stop when requested. 2019-10-10 18:46:29 (6124): VM was NOT successfully terminated. . . . 2019-10-11 07:51:09 (3136): Stopping VM. 2019-10-11 07:51:19 (3136): Error in stop VM for VM: -108 Command: VBoxManage -q controlvm "boinc_0c99027d1c3c10c0" savestate Output: 0%...10%...20%...30%...40%... 2019-10-11 07:51:19 (3136): VM did not stop when requested. 2019-10-11 07:51:19 (3136): VM was NOT successfully terminated. In addition the following line shows that vbox extensions are not installed on your host: 2019-10-10 07:17:41 (13064): Required extension pack not installed, remote desktop not enabled. If you are interested in a better task monitoring those extensions must be installed. See Yeti's checklist: https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=4161&postid=29359 |
Send message Joined: 27 Sep 08 Posts: 831 Credit: 688,871,932 RAC: 126,472 |
I have: 2 at 67hrs 4 at 85hrs 1 at 97hrs 7 at 60hrs Sill have 4 days until the deadline, hopefully they can finish. looks like about 35min per event for my computer there at 130-160 |
Send message Joined: 7 May 08 Posts: 205 Credit: 1,564,905 RAC: 1,765 |
The task finished and got credits - that's BOINC's point of view. Nooo, i partecipate for science not for credits... Your logfile shows that the task started/paused several times and couldn't successfully write it's snapshot I crunch with my notebook, so i turn off it sometimes. But why not snapshot if we use virtualbox?? |
Send message Joined: 7 May 08 Posts: 205 Credit: 1,564,905 RAC: 1,765 |
If you are interested in a better task monitoring those extensions must be installed. Ok, i've installed the latest version of VirtualBox and the latest extensions. But this wu still use 0% of cpu. The precedent wu used 50% of cpu |
Send message Joined: 19 Feb 08 Posts: 708 Credit: 4,336,250 RAC: 0 |
I ended my first 2.0 task in 4 h 44 min and 5 s, with more than 1d, 10 hours, 10 min of CPU time, but no HITS file was produced. Tullio |
©2024 CERN