The Ska runtime environment provides the full runtime resources used for all Chandra “Ska” operations tools. This includes the all aspect related processing, ACA load review tools, the Ska engineering archive, the commanded states database, ESAview tool, etc.
The Ska name was coined in a trailer at TRW during pre-launch integration and test activities. It was originally a lame acronym for Spacekraft und Aspect, but now is just a word that might also refer to an under-appreciated form of music.
For historical reasons, the flight installation has the root path /proj/sot/ska/. The actual root for a Ska runtime environment is determined by the SKA environment variable. The whole environment can be built in a different root, but an existing Ska runtime environment is not movable.
Arch dependent files
This tree holds everything that is architecture and OS-specific, with the exception of Perl, which lives in lib/perl.
FOT Matlab tools depends only on this part of the Ska runtime environment.
arch/
+-- i686-linux_CentOS-5/
+-- x86_64-linux_CentOS-5/
| +-- bin/
| +-- doc/
| +-- etc/
| +-- include/
| +-- lib/
| | +-- python2.7/
| +-- lib64/
| +-- libexec/
| +-- pgplot/
| +-- pkgs.manifest
| +-- sbin/
| +-- share/
| +-- var/
Arch independent scripts
The ipython, perl and python files are just shell launcher scripts that use flt_envs to set the Ska environment and launch the correct arch-specific binary. The flt_envs script depends on the Ska perl to run.
bin/
+-- flt_envs
+-- ipython
+-- perl
+-- python # Launcher script (set environment and launch arch bin python)
+-- ska_envs.csh
+-- ska_envs.sh
+-- sysarch
+-- syspathsubst
+-- ... global Ska application scripts and launchers
Build Ska runtime environment (mostly) from source
build/
+-- i686-linux_CentOS-5/
+-- x86_64-linux_CentOS-5/
Source / binary tarballs for entire Ska runtime environment
pkgs/
+-- ActivePython-2.7.1.4-linux-x86.tar.gz
+-- ActivePython-2.7.1.4-linux-x86_64.tar.gz
+-- Alien-SVN-1.6.12.0.tar.gz
+-- App-Env-0.25.tar.gz
+-- App-Env-ASCDS-0.01.tar.gz
+-- ...
Global includes
include/
+-- Makefile.FLIGHT
+-- SKA_MAKE_DEFAULTS
+-- Task_template/
Global libraries
In reality this only holds Perl 5.8.8, which internally chooses the right architecture where needed.
lib/
+-- perl/
+-- ... <arch independent perl libraries>
+-- i386-linux-thread-multi/
+-- x86_64-linux-gnu-thread-multi/
+-- x86_64-linux-thread-multi/
Project documentation
doc/
+-- cmd_states/
+-- eng_archive/
+-- ...
Project or task data
data/
+-- aca_bgd_mon/
+-- aca_dark_cal/
+-- aca_dark_cal_checker/
+-- aca_egse/
+-- ...
+-- eng_archive/
+-- ...
+-- psmc/
+-- ...
+-- taco/
+-- telem_archive/
Project scripts and files
share/
+-- aca_bgd_mon/
+-- aca_dark_cal/
+-- aca_egse/
+-- ...
Complete Ska runtime environment for development
dev/
+-- arch/
+-- bin/
+-- build/
+-- data/
+-- doc/
+-- idl/
+-- include/
+-- lib/
+-- ops/
+-- pkgs@ -> ../pkgs
+-- share/
+-- www/
Environment setup can be done by eval’ing the $SKA/bin/flt_envs script. This uses sysarch to determine the system architecture.
SKA = /proj/sot/ska
PATH = $SKA/bin : $SKA_ARCH_OS/bin : $PATH
LD_LIBRARY_PATH = $SKA_ARCH_OS/lib : $SKA_ARCH_OS/pgplot : /soft/SYBASE_OCS15/OCS-15_0/lib
PERL5LIB = $SKA/lib/perl : $SKA/lib/perl/lib
The Ska runtime environment is maintained via the skare project. This project consists of a main installer script and a number of configuration files that specify build instructions for each package within Ska. The entire Ska runtime environment can be built from scratch within this project. It requires Python version 2.4 or later to run.
The skare project is maintained under git revision control.
Most regular updates to the HEAD network runtime environment (e.g. updating a component package) are done by placing the new source tarball in /proj/sot/ska/pkgs and updating the skare pkgs.manifest file to reflect the new package. Typically testing is done by first installing to the dev environment /proj/sot/ska/dev with the skare package installer. Once testing is complete the new package is installed to the flight environment with the package installer. The arch, bin, and lib directories are owned and only writable by a management group account aca.
For major updates to the runtime environment, the build is done on a CentOS-5 VM (currently with VMware on Mac). Then the arch-specific directory (e.g. arch/x86_64-linux_CentOS-5) is moved into place after renaming the original. This allows for easy install and quick backout. Note that at this time the Perl part of the environment is largely static and is not part of this process.
The current GRETA network installation follows the same pattern: small updates are done in-place on a per-package basis where possible, large updates are done with a binary install. Many of the compiled packages cannot be built on the standard GRETA network because of the lack of dev RPMs. In this case a binary install is reqiured.
Proposal:
SOT (currently TLA , JC) will maintain primary responsibilty for the skare project and for updates to the content of the Ska runtime environment.
FOT CM will assume ownership and sole write-access for /proj/sot/ska/arch. This directory and contents are henceforth referred to as FOT Ska. All other files in the /proj/sot/ska root will be owned and maintained by SOT.
Changes to FOT Ska are controlled through the FOT Matlab tools control board and will follow all procedures required of actual Matlab code.
FOT CM will track the Ska runtime environment by maintaining a version of the skare project within the FOT version control. Presumably there is no advantage to versioning the actual binary package tarballs.
SOT will maintain a duplicate of the x86_64 HEAD network runtime environment on GRETA in the /proj/sot/ska/sot root (the SOT Ska). This will allow FOT personnel access to the latest versions of SOT tools on chimchim.
SOT will maintain a clone of the skare git repository in /proj/sot/ska/git/skare. The master (aka trunk) branch will reflect the current installation on the HEAD network and the installation in /proj/sot/ska/sot.
On an as-needed basis the SOT Ska will be promoted to the FOT Ska under control of the FOT Matlab tools CCB. Typically this would be driven be a change needed for FOT Matlab tools.
Can FOT Matlab tools be easily configured to use SOT Ska for testing? Or is there a better strategy?