-
Notifications
You must be signed in to change notification settings - Fork 1
Installation
python -m pip install fink_fat
To be able to perform orbit fitting in local mode and recover the trajectories with an orbit, fink-fat use the OrbFit software. We provide a script that will install automatically Orbfit in your home. It will then used Orbfit from the home directory so don't change the Orbfit destination in the script.
To launch the script, you must have the following requirements:
- aria2c, tar, make
Once you have checked your requirement, launch the Orbfit script. It can be found here.
./orbFit_installer.sh
The assocations of alerts perform by fink-fat can generates a large number of trajectories. The local mode of Orbfit is not well suited for this situation due to an important computation time. Fink-fat come with a cluster mode for OrbFit. It use the spark engine to distribute the OrbFit computation over a cluster of machine and drastically speed-up the computation.
We provide a script able to install OrbFit on all the machine of spark cluster. This script can be found here Run the script on the driver of the cluster
./orbFit_cloud.sh
OrbFit is a software that like to generate files, a lot. If you launch the solve_orbit mode of fink-fat with a large number of trajectories, it will generate a really large number of files on your disk. These files are automatically remove by fink-fat but the writing or the reading of these files can slow down the computation. To face of this issue, fink-fat can save these file in a dedicated directory that one can initiating in the fink-fat config file. If this directory is mount on your ram, the writing and the reading of these files will be faster and will increase the life span of your disk (especially, if you have a ssd).
- To create a ram filesystem on ubuntu, refer on this documentation
- On Centos 7, refer to this documentation
We provide also a script to create and mount automatically a ram filesystem on a spark cluster, this script can be found here
./fstab_cloud.sh