Created attachment 185314 [details] v1 patch # QA - portlint OK - poudriere OK in the matrix (10 & 11, vs i386 & amd64) # commit message net-mgmt/riemann: update 0.2.12 to 0.2.14 upstream changes: - speed improvements - support microsecond resolution internally and in GPB wire format - kafka, netuitive, telegraph, msteams plugins added - elasticsearch, telegram, pushover plugins updated full details at https://github.com/riemann/riemann/blob/master/CHANGES.markdown port changes: - move pids to /var/run/riemann/ so we have a pid for daemon(8) and riemann - improve SIGHUP handling with new pid dir arrangement - ensure daemon(8) invocation has matching title in ps(1) output - bump lowest JAVA_VERSION to 1.8+ missed from last release
Created attachment 185316 [details] v2 valid svn diff
Hi Dave. Thanks for your work. One doubt in the Makefile. > RIEMANN_USER?= ${PORTNAME} > RIEMANN_GROUP?= ${PORTNAME} > > .if ${RIEMANN_USER} == "${PORTNAME}" > USERS= ${PORTNAME} > .endif > .if ${RIEMANN_GROUP} == "${PORTNAME}" > GROUPS= ${PORTNAME} > .endif Are there some reason for this? Regards.
sorry for the delay, I've been on vacation. TBH this section would have been copied from whatever JVM port I originally used as a template - this was my first port. I don't see any issue with removing it and just using riemann/riemann as std group and port. Is there an alternative solution you'd recommend?
(In reply to Dave Cottlehuber from comment #3) Hi Dave. So I'll just simplify this. Thanks for the feedback.
Created attachment 185953 [details] riemann-simple.patch Take a look if this patch is ok for you, the idea is to keep Makefile simple because riemann is not a MASTER_PORT.
Created attachment 185965 [details] v3 patch with Danilos' changes and even simpler pid management Thanks Danilo. I like your new uid/gid approach, its much more readable. For pidfile management, just a clarification: we need to be able to SIGHUP the JVM directly, to allow reloading the configuration, as daemon(8) doesn't propagate this signal down. Now that I am using the pkill approach to identify the correct JVM process, limited by group, we can just use a single pidfile for daemon(8). Does this look ok to you? "works for me here"..
A commit references this bug: Author: dbaio Date: Fri Sep 1 00:07:49 UTC 2017 New revision: 449053 URL: https://svnweb.freebsd.org/changeset/ports/449053 Log: net-mgmt/riemann: Update 0.2.14 Bump lowest JAVA_VERSION to 1.8+ missed from the last release and simplify port because it's not a master port. PR: 221436 Submitted by: Dave Cottlehuber <dch@skunkwerks.at> (maintainer) Changes: head/net-mgmt/riemann/Makefile head/net-mgmt/riemann/distinfo head/net-mgmt/riemann/files/riemann.in head/net-mgmt/riemann/pkg-plist
Committed, thanks!