Prefix Portage on Solaris: setup script

rafo

Working in Solaris environments can be uncomfortable if you are used to Linux. It is Unix, but many commands behave slightly different, or lack the options that they "should" have :wink:

Gentoo can be the remedy though! This webpage describes how to install a Gentoo-like environment in an ordinary Solaris user account: http://www.gentoo.org/proj/en/gentoo-alt/prefix/bootstrap-solaris.xml. I managed to follow it although I had to find some workarounds. When doing it for a second time I decided to record the steps in a script.

The script is here: http://wb748077.bahnhofbredband.se/prefix-gentoo/. Try it out if you like; no guarantees.

rafo · Posted: Sat Jun 23, 2007 7:30 pm Post subject:

I have uploaded an improved version of the script: It now runs on both Solaris/x86 and Solaris/SPARC. It can be downloaded from the same link as before.

vandien · Tux's lil' helper Joined: 03 May 2006 Posts: 137

Thanks for the script, I'm running it now on a v215. Only one problem came up that I had to fix which was linking libucb.so.1 to /usr/lib for the initial ncurses emerge (dependency of bash). maybe there's a better way to fix this...

rafo · Posted: Fri Jul 06, 2007 9:25 am Post subject:

Hi vandien,

your fix indicates that you have root access. It should however be possible to do the entire build process as an unprivileged user. I will try and see if your problem can be reproduced over here (the system that I have access to identifies itself as SunOS 5.10 Generic_118833-33 sun4v sparc SUNW,Sun-Fire-T200).

Also, since I submitted my script in this forum I have become aware of another initiative towards setting up "Prefix Gentoo". Here is the link: http://sourceforge.net/projects/prefix-launcher/.

rafo · Posted: Fri Jul 06, 2007 7:50 pm Post subject:

Vandien, I wasn't able to reproduce your "ncurses" problem. I ran the script on both a Solaris/x86 system and a Solaris/SPARC system, and ncurses was emerged just fine.

However, later on I was stuck at the point of Code Listing 1.8 where gcc would not build. The details are in this bug report: https://bugs.gentoo.org/show_bug.cgi?id=184437.

vandien · Tux's lil' helper Joined: 03 May 2006 Posts: 137

hey, thanks for the response. I'm not sure what my problem was caused by. I did finish the install though after that error and everything seemed to work as expected. However I wasn't totally sure how to proceed after that. I noticed the portage tree was a bit mangled (couldn't emerge apache or php, which was the end result I was looking for) and attempts to add the ebuilds with an overlay failed. One thing I forgot to do was apply the 10_recommended updates after a clean install of solaris. I'm new to solaris, so still learning a lot. I'll probably try using your script again (or the other link you posted) soon.

For now I'm using coolstack to test an AMP server.

rafo · Posted: Sat Jul 07, 2007 10:34 pm Post subject:

I think I have found out what causes https://bugs.gentoo.org/show_bug.cgi?id=184437, and I uploaded a modified script that has a workaround for that bug (same location as before). But alas, the new script fails on another problem at step 1.19 (https://bugs.gentoo.org/show_bug.cgi?id=184543).

I will be away from Solaris machines for the next 10 days or so. Anyone, feel free to tinker with the script :wink:

rafo · Posted: Tue Jul 24, 2007 7:24 am Post subject:

Good news, the prefix portage tree has evolved and right now the 2007-07-07 version of the script runs successfully on Solaris/x86. I will go on to try it out on Solaris/SPARC now.

rafo · Posted: Thu Aug 30, 2007 7:07 pm Post subject:

I have uploaded a new version of the script, to the same location as before. It is more or less in line with the August 15 version of the Gentoo Prefixed Portage Bootstrap Process for Solaris, but there are some workarounds.

Each workaround was introduced at some point in time to make the script succeed unattended. It may well be that some of them can be undone by now. Or, new workarounds may be needed if your Solarises differ from mine...

I added timestamps to the console output. My final timestamp for Solaris/x86 says 9:28 hours, so running the script is a typical Gentooish overnight operation