Difference between revisions of "SAC TODO List"

From OSGeo
Jump to navigation Jump to search
 
(cleanup, new link to trac tickets)
 
(7 intermediate revisions by 3 users not shown)
Line 1: Line 1:
= Short Term =  
+
'''This page is stale, todo items now primarily kept in [http://trac.osgeo.org/osgeo/query?status=assigned&status=new&status=reopened&component=Systems+Admin&order=priority Trac]'''
* Get LDAP user authentication working on all configured servers with the possible exception of the LDAP server itself.
+
 
 +
== Short Term =
 
* Share network user directories between systems.  
 
* Share network user directories between systems.  
 
* Define [[SAC:Security Groups Policy]]
 
* Define [[SAC:Security Groups Policy]]
* Come up with a procedure for creating new users, and altering user groups reasonably conveniently.  
+
* Admit more volunteers to SAC.
* Setup SVN server(s).  
+
* Build a [[Membership Application]].
  
= Longer Term =
+
== Longer Term ==
 +
* Setup buildbot master system.
 +
* Define backup strategy and put into operation.
  
* Setup/migrate mediawiki.  
+
== Issues to bear in mind ==
* Setup buildbot master system.
+
 
 +
(This is content moved from the [[Infrastructure Working Group]] page after SAC was formed)
 +
 
 +
=== Knowledge Transfer ===
 +
This group is going to need to document procedures, locations of things, and so forth quite carefully so that stuff doesn't get completely forgotten, and so that we aren't too limited to one person to fix things.  This will require dicipline but will be important.
 +
 
 +
=== Permissions Management ===
 +
 
 +
We will need a way of managing the permissions level of various OSGeo members.  It is planned that an LDAP server will be used to handle the user database (a distinct list from those at CollabNet for now). But does this address how we delegate permissions to different people?  How do we decide who has login permissions on test systems for instance, or who has update permissions in a given Plone instance?
 +
 
 +
Some categories of permissions we need to address:
 +
* Some systems would only provide login access to the system administration working group.  These systems we would hope could be very stable.  
 +
* Login on some systems, such as test systems would need to be very widely available, essentially to all OSGeo project developers for testing.  Presumably normal developers would not require root access, but we could still assume these systems are likely to be subject to damage and need to be fairly easily recreated.
 +
* We may also want "Project" systems, roughly owned by a single project with appropriate project staff having full (ie. root) permissions.  Can we provide some sort of virtualization to offer roughly this to people?  Perhaps "chrooted jails"?  This might be the sort of place that projects would have automated scripts running to create nightly snapshots, re-create projects web pages or whatever.
 +
 
 +
[[Category:Infrastructure]]

Latest revision as of 10:38, 17 September 2011

This page is stale, todo items now primarily kept in Trac

Short Term

Longer Term

  • Setup buildbot master system.
  • Define backup strategy and put into operation.

Issues to bear in mind

(This is content moved from the Infrastructure Working Group page after SAC was formed)

Knowledge Transfer

This group is going to need to document procedures, locations of things, and so forth quite carefully so that stuff doesn't get completely forgotten, and so that we aren't too limited to one person to fix things. This will require dicipline but will be important.

Permissions Management

We will need a way of managing the permissions level of various OSGeo members. It is planned that an LDAP server will be used to handle the user database (a distinct list from those at CollabNet for now). But does this address how we delegate permissions to different people? How do we decide who has login permissions on test systems for instance, or who has update permissions in a given Plone instance?

Some categories of permissions we need to address:

  • Some systems would only provide login access to the system administration working group. These systems we would hope could be very stable.
  • Login on some systems, such as test systems would need to be very widely available, essentially to all OSGeo project developers for testing. Presumably normal developers would not require root access, but we could still assume these systems are likely to be subject to damage and need to be fairly easily recreated.
  • We may also want "Project" systems, roughly owned by a single project with appropriate project staff having full (ie. root) permissions. Can we provide some sort of virtualization to offer roughly this to people? Perhaps "chrooted jails"? This might be the sort of place that projects would have automated scripts running to create nightly snapshots, re-create projects web pages or whatever.