Features :
DONE :
- Easier and documented install procedure
- Use of Scons
- A general and unique configuration file can be use to configure the whole application in mono-node (multi node to be done)
- Data/log/binary can be separated on different partitions
- TCP port can be changed
- MySQL password management
- automated procedure for PT1.1 Object data loading
- automated procedure for PT1.1 meta generation
TODO :
- could be easilly generalized to PT1.1 Source table, and to other data-set
- runTests.py doesn't work. It could be completed with other tests related to PT1.1.
- Commit in the master branch
Maintenance contract :
The workflow could be the next :
The french team can adapt and validate this procedure for next Qserv official release, with the help of Qserv developpers. Furthermore, the french team proposes to insure maintenance on this installation procedure, for next Qserv official release, on a bi-node architecture. So, if system administrators encounter problems during installation of this release, the french team will help them to install it and to run tests.
For Qserv futures releases, the french team is very interested to continue to maintain and integrate this installation procedure, but they'll first do a serious study in order to define if this commitment can be respected depending on their manpower, and on complexity of source code evolution.
Remark : The quality of the procedure will be leveraged it will be easy to maintain.
- DSH is going to send a mail explaining of to load qserv meta. FJS and EMH are going to test this, and then try to automate it.
mysql> select * from Object where objectid=402395485975435; Empty set (4.71 sec) mysql> select * from Object where objectid like '402395485975435'; work fine
- DWG is working on a new version of qserv which will be more modular and will allow FJS and EMH to propose their help on qserv core.