LHC@home

From BOINC client messages

Fr 11 Aug 2017 16:40:56 CEST | LHC@home | Entry in app_config.xml for app 'ATLAS', plan class 'native_mt' doesn't match any app versions

According to this tread it should be on this project server, but it isn't.
What now?

<edit>
I got a vbox ATLAS instead that failed due to the lower RAM setting for the expected native app.
</edit>

The client has no knowledge of the app version until it has received at least one task using that app version.

Trying to run the native ATLAS app with Debian Jessie.

I successfully compiled CVMFS (cvmfs_config probe shows all "yes") and singularity (at least i think so, just tested it with singularity --help). Problem is that the Debian machine dont get any native ATLAS tasks. Just the normal vbox ones.

I have "run test applications?" enabled.

Is there any way to force the server to send native tasks instead of the vbox tasks?

Problem is that the Debian machine dont get any native ATLAS tasks. Just the normal vbox ones.

I have "run test applications?" enabled.

Is there any way to force the server to send native tasks instead of the vbox tasks?

I am not getting any on two Ubuntu machines that seem to be set up correctly also. I expect they are still fixing the problems from the last run, and don't have any available.
There will possibly be another release next week rather than a late release this week I would think.

As mentioned on the original news item, only RHEL versions 6 and 7 (and variants) are supported for the native app at the moment. Once we iron out all the remaining issues we will open it up to other linux versions.

Since this is just another version of the ATLAS app, the WU queue is the same for native and vbox WU. The BOINC scheduler decides for each client request whether to send a native or vbox version but the WU is the same - so if there are ATLAS WU in the queue then it's possible to get native WU if you have the right linux version.

In my own testing I found that the only way to force native tasks is to remove virtualbox.

cvmfs_config probe
Probing /cvmfs/atlas.cern.ch... Failed!
Probing /cvmfs/atlas-condb.cern.ch... Failed!
Probing /cvmfs/grid.cern.ch... Failed!

No ping - "unknown host atlas.cern.ch"

Do I need to change something in the settings?

cvmfs_config probe
Probing /cvmfs/atlas.cern.ch... Failed!
Probing /cvmfs/atlas-condb.cern.ch... Failed!
Probing /cvmfs/grid.cern.ch... Failed!

No ping - "unknown host atlas.cern.ch"

Do I need to change something in the settings?

Have you used this script for installation?:
http://atlasathome.cern.ch/boinc_conf/install_cvmfs_sin.sh

Are you using a firewall or proxy to connect to the internet?
If your are using a firewall, you have to open the correct ports. These are listed in the FAQ http://lhcathome.web.cern.ch/test4theory/my-firewall-complaining-which-ports-does-project-use. For CVMFS its port 3125.

Yes, I used the installation script on my SLC6 host
Probe worked immediately after the installation.
It worked before, when I tried to manually connect Ubuntu at the first start of the native application

But when I received the first tasks from the LHC, all ended with an error

There is no proxy or firewall

The problem in resolving the addresses for hosts from the configuration file
CVMFS_REPOSITORIES=atlas.cern.ch,atlas-condb.cern.ch,grid.cern.ch

>nslookup atlas.cern.ch 8.8.8.8
Server: google-public-dns-a.google.com
Address: 8.8.8.8

*** google-public-dns-a.google.com не удалось найти atlas.cern.ch: Non-existent domain

Yes, I used the installation script on my SLC6 host
Probe worked immediately after the installation.
It worked before, when I tried to manually connect Ubuntu at the first start of the native application

But when I received the first tasks from the LHC, all ended with an error

try: sudo service autofs start
and then: cvmfs_config probe

One time SL6.9 was probing OK.

Now also say failed.

Warning: SElinux enabled, but cache directory (/scratch/cvmfs) labeled 'default_t' instead of 'cvmfs_cache_t'

Have used debugging from:

http://cernvm.cern.ch/portal/filesystem/debugmount

but no success.

Boinc from atlas_conf is Vers.7.5.1.
FYI under help it shows 6.10.58.

Have for testing now running one WCG task. This seem ok.

As mentioned on the original news item, only RHEL versions 6 and 7 (and variants) are supported for the native app at the moment. Once we iron out all the remaining issues we will open it up to other linux versions.

yes. ok, then i most likely have missunderstood this post

In my own testing I found that the only way to force native tasks is to remove virtualbox.

Althoug i have CentOS running now on one of my computers, the server sent a Vbox task. So i removed Virtualbox and now it gets native ones.
And as far as i can tell, they are runnnig without any problems.

Once its out of beta, maybe an own application checkbox (like CMS, LHCb...) on the preferences site for the native app would be worth considering.

running cvmfs_config reload together with
CVMFS_HTTP_PROXY="DIRECT" in default.local in etc\cvmfs have success.

sudo cvmfs probe and sudo cvmfs chksetup are OK now.

singularity is installed.

Boinc from 7.5.1 changed to 7.2.42.

Boinc from 7.5.1 changed to 7.2.42.

Have a problem to install 7.2.42, so went back to 7.5.1 from atlasathome.cern.ch/boinc_conf.

After connect to LHCatHome, the message of Boinc is Virtualbox is not installed.

Will install it now, but it is not necessary.

https://lhcathome.cern.ch/lhcathome/show_host_detail.php?hostid=10495075

probe and checksetup of cvmfs are failing at the moment.
Is this because of the nightly update?

probe and checksetup of cvmfs are failing at the moment.
Is this because of the nightly update?

It is working for me flawlessly (with some warnings in checksetup). Also the tasks are runnning without a problem (if there are some :-) ).
Update? I still get 2.50 version. Or what update do you mean?

Does checksetup show you something why it is not working?

Is it possible for you to open your Computerlist?

Is it possible for you to open your Computerlist?

should be visible now

Thank you, have one SL6.9 and get this message:

Warning: SElinux enabled, but cache directory (/scratch/cvmfs) labeled 'default_t' instead of 'cvmfs_cache_t'

It seem a conflict with cernvmfs and the scratch of cernvmfs.
Will install cvmfs once again.

Have in the moment WCG tasks running and wait for finish.

SL69 after new installation of install_cvmfs_sin.sh now with:
Probing /cvmfs/atlas.cern.ch... atlas-condb.cern.ch grid.cern.ch all three are OK.
chksetup OK.

Boincmanager(7.5.1) with
No tasks are available for ATLAS Simulation and
Message from server: Virtualbox is not installed.

Any idea?

Is the work from atlasathome-Server?

Edit: Have ONE Cpu and ONE Job. Is 5GByte RAM ok?

This are the last lines after Installation:

VERSION PID UPTIME(M) MEM(K) REVISION EXPIRES(M) NOCATALOGS CACHEUSE(K) CACHEMAX(K) NOFDUSE NOFDMAX NOIOERR NOOPEN HITRATE(%) RX(K) SPEED(K/S) HOST PROXY ONLINE
2.3.5.0 5673 0 28828 26954 14 1 242870 4194305 0 65024 0 0 n/a 4159 1 http://cvmfs.racf.bnl.gov/cvmfs/atlas.cern.ch http://206.167.181.94:3128 1
cvmfs is installed
2.2.99
singularity is successfully installed..
[root@APU10S Downloads]# cvmfs_config chksetup
OK
[root@APU10S Downloads]# cvmfs_config probe
Probing /cvmfs/atlas.cern.ch... OK
Probing /cvmfs/atlas-condb.cern.ch... OK
Probing /cvmfs/grid.cern.ch... OK