Hi all, a question regarding the new IPS, that has been gnawing on me for a while now.
First some context: I have a TON (and I mean HUNDREDS) of System V packages that currently automate and completely encapsulate various system configuration tasks. For example, I have a package that automatically configures syslog.conf on a firewall to log to a specific log file; a package that automatically modifies $PKG_INSTALL_ROOT/etc/nsswitch.conf to use DNS; a package that automatically adds user "oracle" and groups "dba" and "oinstall" to a system; a package to configure the system's $PKG_INSTALL_ROOT/etc/inet/hostname to configure "127.0.0.1" to also be a "localhost"; and so on. In other words, I have automated a whole bunch of system administration tasks, thereby making them consistently repeatable via pkgadd(1M), as per the Capability Maturity Model (CMM) level 2 and above. Now to my question: from what I understood of IPS so far, scripting is, at least currently not supported and not implemented in IPS. How do you plan to support system engineers and sysadmins with regards to automation and encapsulation with IPS? So far, pkgadd(1M) has offered perhaps not perfect, but at the very least consistent mechanism for modifying the system configuration via class action scripts and pre- and postremove and pre- and postinstall mechanisms. Consistency and repeatability is key here. Some mechanism of encapsulation will be needed. There is currently an "action" defined for installing drivers, but little else in the way of delivering or modifying configuration files on the system, or thousands of systems. What, if anything, is planned? -- This message posted from opensolaris.org
