Sowmini Varadhan wrote: >> http://opensolaris.org/os/project/caiman/auto_install/system_config_design.txt. >> >> Please review and give your feedback. >> >> Thanks, >> Sundar >> > > Hi, > > thanks for sharing this doc- I'm cc-int networking-discuss, where we are also > looking at various > improvements to Configuration..some questions whose answers are not clear to > me from > reading the document above.. > > Regarding option (c) which seems to be the favored approach- > most of solaris cannot be configured using smf today, e.g., things like > user account, root passwd etc are not smf properties, but rather, they > are stored in flat files. Taking the user account as an example, > one adds a new user to the live system using useradd(1m) and this results > in modifications to several files in /etc. > > How does the proposed design envision this happening with smf? > Section 5.1 discusses how we are thinking of approaching the different types of configuration. For example, for a configuration like root password, the install may have a special SMF service to handle the setup. > Moreover, there's the question of input validation. An application like > useradd(1m) is capable of doing semantic validation of the input data > that is specific to the task being performed. How can a general framework > like SMF provide the same level of input validation? > The install tools can do the syntactic validation when the SMF profile is added by the user. The SMF could do some validation to check whether the properties specified in the profile are defined in the corresponding service. However, it is the responsibility of the the component that handle a specific configuration to make sure that the property value makes sense.
- Sundar > --Sowmini > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/caiman-discuss/attachments/20091001/b9d17580/attachment.html>
