Changelog for s6-rc. In 0.5.0.0 ---------- - Adaptation to skalibs-2.8.0.0 and s6-2.8.0.0 - Bugfixes. - s6-rc-compile doesn't support -u and -g anymore; the user who can run s6-rc on a database is the user who owns the supervision tree. In 0.4.1.0 ---------- - Adaptation to skalibs-2.7.0.0 and s6-2.7.2.0. - Support for max-death-tally and down-signal files in a longrun definition directory. In 0.4.0.1 ---------- - Improved portability to (guess what) OpenBSD In 0.4.0.0 ---------- - s6-rc-compile now accepts several producers in a consumer-for file. All the producers for a service will write to the same pipe. In other words: pipelines are now funnels. - The pipeline-name file, defining the name of a bundle containing all the services in a pipeline, must now be place in the definition directory of the last consumer in the pipeline, instead of the definition directory of the first producer (since the latter is not unique anymore). - All service databases must be edited and recompiled. - New tool: s6-rc-format-upgrade, to handle the upgrade to the new compiled format. In 0.3.0.0 ---------- - s6-rc-init now takes a -p option that adds a prefix to the symbolic links created in the scandir to manage longruns. This allows the user to use the same scandir for several s6-rc databases and live directories without risking conflicts, if the prefixes are unique enough. - s6-rc-update does not use random number generation anymore, so it does not block at boot time. - Bugfixes. Despite this being a major version release, the compiled service databases did not change formats; you do not need to recompile your service databases. In 0.2.1.2 ---------- - Bugfix release. - s6-rc's verbosity has been adjusted. (-v2 is less verbose, -v3 is what the previous -v2 was, -v4 is still extremely verbose.) In 0.2.1.1 ---------- - Bugfix release. In 0.2.1.0 ---------- - Timeouts for oneshots have been increased to 30 seconds. - s6-rc now sends an explicit "s6-svc -d" to a longrun when an up transition fails. - s6-rc now kills subprocesses waiting on longrun transitions when it receives a SIGINT or a SIGTERM. + Those two changes combined ensure that a user can interrupt a transition that is blocked in a "service is not-ready" limbo, and bring back the service to a known "down" state. In 0.2.0.1 ---------- - Bugfix release. - Support for timeout-kill (in s6-2.5.1.0). In 0.2.0.0 ---------- - Added this NEWS file. :) - Major types overhaul to make them more POSIXly correct: compatibility with skalibs-2.5.0.0. - s6-2.5.0.0 now allows longruns to report permanent failures (when ./finish exits 125). s6-rc should pick that up automatically and fail a longrun transition when it's the case. - s6-rc commands taking a lock on the live directory or the service database now fail instantly if they're already locked. A -b option has been added to give them the "block and wait for the lock to be released" behaviour back. - Easier external interaction with the hidden s6rc-fdholder service: the list of pipes to autofill a new fdholder with is now taken from the data/autofilled file (previously: was a hardcoded command line in the s6rc-fdholder run script).