Getting started as a cAos project buildnode The cAos project makes its un-completed build SRPM candidate and Hint queue externally visible, and so also its build-results, in a form which permit anonymous and automated access to these build sources. This outline documents the process (mixed manual and scripted) by which a build node may be set up, and process those build queue candidates. There are 'trusted' and 'untrusted' nodes, with a difference that the binary build packages and build history from a 'trusted' node are then capable of being uploaded back to the cAos project servers, and further considered for inclusion in the 'Crazy' incoming testing queue. There is a post-package QA, review, integration and release to 'Crazy' post-process which occurs before before it would be released to end user access. But those matters are for after one has a working buildnode setup in place. Easy decisions are: 1. What build architecture (x86 is most common) 2. What target cAos product (caos-1 is the current lead) 3. What build host base (again, caos-1 is the current lead) Obviously more processor speed, more ram, and a big hard drive will yield more packages more quickly. TBD: - getting a build chroot set up =- reference ftp://ftp.owlriver.com/pub/local/ORC/buildfarm/ORCyum-chroot - variant needed TBD: - getting local binary packages (ftp mirroring of a hot archive) for local Hint insertion more quickly TBD: - getting build candidates, hints into local host TBD: - setting up a localized ORCbuildit ftp://ftp.owlriver.com:/pub/local/ORC/buildfarm/ TBD: - common problems; getting stuck, watching for /proc infringement TBD: - Returning results upstream; trusted hosts discussion TBD: - Build error evaluation; troubleshooting common build problems; local fork and fix process outline; filing against SRPM with Adopter; sending patches upstream -------------- Initial - RPH - 040128 couch:~/caos/getting_started_buildnode.txt external: http://www.herrold.com/caos/getting_started_buildnode.txt ----------------