Message boards : News : New ATLAS app version released for Linux hosts
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Profile Olivier Fehr
Avatar

Send message
Joined: 1 Jun 17
Posts: 9
Credit: 964,242
RAC: 0
Message 31937 - Posted: 13 Aug 2017, 7:51:01 UTC - in response to Message 31935.  

@Ivan, Jim1348

It's a bit OT here, therefore only some very short remarks.

CERN's vbox apps can get a huge benefit from a local squid.
CMS benefits most, then ATLAS and LHCb, Theory less but sufficient.
Up to 95 % regarding the request hits and 50 % regarding the byte hits.
The setup needs some special tuning compared to the standard setup and should get some assist from iptables.

An "old" computer with a core 2 duo, 1-2 GB RAM (250-350 MB for squid), 30-80 GB free disk space is far enough.
And there is also a benefit for fast DSL lines as the squid avoids latency.


Thanks for that insight. I going to try this, since I have a squid running already.
ID: 31937 · Report as offensive     Reply Quote
ivan
Volunteer moderator
Project tester
Volunteer developer
Volunteer tester
Project scientist
Avatar

Send message
Joined: 29 Aug 05
Posts: 488
Credit: 3,834,148
RAC: 3,511
Message 31939 - Posted: 13 Aug 2017, 10:18:36 UTC - in response to Message 31935.  

@Ivan, Jim1348

It's a bit OT here, therefore only some very short remarks.

CERN's vbox apps can get a huge benefit from a local squid.
CMS benefits most, then ATLAS and LHCb, Theory less but sufficient.
Up to 95 % regarding the request hits and 50 % regarding the byte hits.
The setup needs some special tuning compared to the standard setup and should get some assist from iptables.

An "old" computer with a core 2 duo, 1-2 GB RAM (250-350 MB for squid), 30-80 GB free disk space is far enough.
And there is also a benefit for fast DSL lines as the squid avoids latency.

Yes, you are right there, I wasn't considering that use case. Every running instance (of the CMS app at least) has its own cvmfs built-in to the VM. A local squid would definitely help if you run several instances. Do you have a guide as to how to re-direct the VMs to a local squid instead of, e.g. lhchomeproxy1.cern.ch:3404 which is baked-in to the CMS vdi?
ID: 31939 · Report as offensive     Reply Quote
tullio

Send message
Joined: 19 Feb 08
Posts: 512
Credit: 2,176,190
RAC: 744
Message 31941 - Posted: 13 Aug 2017, 13:11:37 UTC
Last modified: 13 Aug 2017, 13:15:00 UTC

I have two Linux boxes with SSE2 capability. Yet one of them, the E-450 CPU is receiving SixTrack Tasks with no sse2 or pni, resulting in very long running times. One is going to be late. The preceding task was completed and validated in 399,883.43 s. This is computer 10492633. The other computer,10492932, an Opteron 1210 CPU, is running Einstein@home CPU tasks with AVX.
Tullio
ID: 31941 · Report as offensive     Reply Quote
Jim1348

Send message
Joined: 15 Nov 14
Posts: 200
Credit: 5,819,644
RAC: 3,430
Message 31976 - Posted: 16 Aug 2017, 20:27:43 UTC

As I have asked on the Wishlist, there should be a listing of the ATLAS beta tasks on the Project Status page, so we know what is going on (or not).
ID: 31976 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 472
Credit: 14,911,824
RAC: 24,971
Message 32626 - Posted: 4 Oct 2017, 11:41:38 UTC
Last modified: 4 Oct 2017, 11:42:13 UTC

wan't to test OpenSuse 42.2 (Leap) as native-app.

Up to yum the commands are running

yum -y install cvmfs-keys cvmfs cvmfs-init-scripts

http://cvmrepo.web.cern.ch/cvmrepo/yum/cvmfs/EL/42.2/x86_64/repodata/repomd.xml: [Errno 14] HTTP Error 404 : http://cvmrepo.web.cern.ch/cvmrepo/yum/cvmfs/EL/42.2/x86_64/repodata/repomd.xml
Trying other mirror.
Error: Cannot retrieve repository metadata (repomd.xml) for repository: cernvm. Please verify its path and try again

There is a cvmrepo.web.cern.ch/cvmrepo/yum/cvmfs/x86_64

What is the correct cvmfs.rpm?
ID: 32626 · Report as offensive     Reply Quote
AlphaC

Send message
Joined: 6 Sep 13
Posts: 2
Credit: 463,908
RAC: 1,231
Message 35410 - Posted: 1 Jun 2018, 21:15:38 UTC - in response to Message 31904.  
Last modified: 1 Jun 2018, 21:16:54 UTC

I am just curious why the install script still uses yum exclusively instead of dnf?

dnf has been made the default installer since Fedora 22. I suppose since the installer was oriented to CentOS 7?
ID: 35410 · Report as offensive     Reply Quote
Profile Joe's Collider

Send message
Joined: 13 May 12
Posts: 2
Credit: 66,287
RAC: 113
Message 35489 - Posted: 12 Jun 2018, 5:42:34 UTC

I don't know if this helps much since I don't know what your code uses, and you might already be doing this now...
Reading the original post, reminded me about LSB (Linux Standard Base), and limiting your code to calling only what's in LSB that would give you a larger audience.
Here's information on LSB: https://refspecs.linuxfoundation.org/lsb.shtml
...if you're already doing this, ignore this reply.

...under the same thoughts, this may also help: https://en.wikipedia.org/wiki/C_standard_library
ID: 35489 · Report as offensive     Reply Quote
Previous · 1 · 2

Message boards : News : New ATLAS app version released for Linux hosts


©2018 CERN