CERN had its OpenDays on September 14 and 15. As the LHC is in Long Shutdown 2 (LS2) for upgrades until early 2021, this was a good possibility for CERN to present itself and its work to the public.
Both days drew huge crowds and lines for underground visits were long – at one point waiting times for ATLAS visits were 3 hours.
I arrived on Sunday, September 15 shortly before 10 a.m. and after getting my wrist band at the check-in tent went straight for transport to remote site – I already know part of the Meyrin site, and Atlas was already overcrowded so I went to the bus stop in search of Bus F, to go to the CMS Experiment site. Unfortunately, I couldn’t find this bus, so I decided to jump on the one going to the LHCb site. Good choice!
I have been using BOINC software to participate in scientific computing projects for around four years and contributed to several projects such as Einstein@Home, SETI@Home, Asteroids@Home and my personal favourite LHC@Home.
Starting with getting LHC@Home projects directly from LHC@Home, I switched to a pool with Gridcoin. I am now switching back and let my boxes crunch exclusively for LHC@Home.
My Boinc four clients now use a local SQUID proxy especially configured for LHC@Home and CERNVM-FS. While the number of machines probably does not do much to cut down on network usage, it’s something I tried some years ago and had abonded it. Apparently, LHC@HOme is now recommending you run a local proxy if you have several crunchers in your network.
As per documentation “CernVM 4 is a virtual machine image based on Scientific Linux 7 combined with a custom, virtualization-friendly Linux kernel”. It’s base image is very small, which means currently around 20MB. The rest of the OS and applications is downloaded on demand via CernVM-FS.
I learned a lot in the recent CernVM/CernVM-FS workshop at CERN in Geneva (actually in the part in France). It offered interesting approaches and insights in how to work with bigdata in complex environments, where almost every user has her own requirements and software setup.
The current CernVM image can be downloaded from here. There are images for different virtual environments available. I chose the VirtualBox version as I have worked with VirtualBox for quite a while using it with BOINC and LHC@Home, but you can chose another environment, for example on AWS, Azure or Docker image.
So far so good. The next step is getting an CERN account. This is needed to access CernVM Online to create a CernVM Context. Once the context is created online, the CernVM on the desktop needs to be paired with the online context. This will automatically configure the CernVM for your needs.
This is simple enough. A few difficulties arise, however. First, for the un-initiated like me, identifying the resources – i.e. the software and data – needed to work with CernVM. Is quite a challenge. Second, I to identify the account and determining the permissions needed is also challenging. While you can register for the Cern website with, say, your Google account, this will give you access to some resources but apparently not CernVM Online. A Cern light-weight account gives access to CernVM but so far, while I got CernVM running and associated it successfully with a online context, I so far get ‘access denied’ on underlying resources.
While CernVM is still work in progress, it inspired me to look a bit closer into VirtualBox and its possibilities and I am currently in the process of moving the development environment of my sun.spaceobservatory.ru (a.k.a sun.ofehr.space) website onto a VM.