Please find attached an updated ebuild and init script for bacula version 1.36.1. Reproducible: Always Steps to Reproduce: 1. 2. 3.
Created attachment 46340 [details] ebuild file
*** Bug 75455 has been marked as a duplicate of this bug. ***
Created attachment 50432 [details, diff] bacula_log.diff Patch for logging to /var/log/bacula instead of /var/bacula/log.
Comment on attachment 50432 [details, diff] bacula_log.diff The patch also copies a configuration file for logrotate to /etc/logrotate.d
While the name and location of the log file may appear to have been an oversight, it is in fact the intention of the designer. Patching the software in this manner breaks it in the sense that it no longer conforms to the published documentation. Such a patch, in later releases, will also have to deal with the logwatch script now in cvs.
I dont't think I'm an FHS fanatic, but I do like all logfiles going to /var/log (as with configuration files in /etc). However, this is only a suggestion. OTOH, I strongly agree with you (Scott) regarding the documentation etc. If done right, patching the logfile location must imply patching man pages and other documentation as well (what I failed to do...). BTW, I took a quick look into bacula.pdf and found that the paths in several examples are already out of sync - even without my patch ;)
Created attachment 50534 [details] Revised ebuild corrects bug 75020 and adds logrotate script This revised ebuild corrects the sandbox violation noted in bug 75020 with a patch to the configure script and adds in the logrotate script previously missing. Requires 1.36.1-cdrecord-configure.patch
Created attachment 50535 [details, diff] patch for configure script to correct sandbox violation Removes call to cdrecord in configure script to address bug 75020
Wenn installing the documentation, the ebuild calls [...] | dohtml -r ${S}/doc/html-manual doc/home-page | chown -R root:root ${D}/usr/share/doc/${PF} | chmod -R 644 ${D}/usr/share/doc/${PF}/* This leaves the html docs unreachable as the 'execute' permissions are removed from all directories under /usr/share/doc/${PF}. The chmod call should be removed, dodoc and dohtml already install regular files with 644 permissions. The call to chown seems to be obsolete, at least if you call emerge as root.
Created attachment 50656 [details] revised ebuild corrects documentation bug remove chown and chmode on documentation files.
*** Bug 83610 has been marked as a duplicate of this bug. ***
Created attachment 52811 [details, diff] 1.36.2-cdrecord-configure.patch
Created attachment 52812 [details] bacula-1.36.2.ebuild
Created attachment 52813 [details] digest-bacula-1.36.2
Hello, I have uploadet the ebuild from the bacula sf page, please add the ebuild fast to the main portage. Thanks Daniel Heule
*** Bug 84501 has been marked as a duplicate of this bug. ***
*** Bug 75519 has been marked as a duplicate of this bug. ***
Please review my patch file and ebuild a bit closer. If you do not use the proper patch for 1.36.2 the problem will NOT go away. I had to grep through other files and directories to come up with a properly working solution. In addition, my ebuild file adds the ability to install a console on the client, or any other machine. This is important for those situations where the "server" may be headless (as is my case), or the server is not to be used for administrative logins. Thomas Cameron
Created attachment 52941 [details, diff] Properly working CDRECORD patch for 1.36.2 This patch works CORRECTLY for version 1.36.2, and the patch from 1.36.1 does not.
Created attachment 52942 [details] Add the ability to install the console apps This ebuild has the added ability to control the build and install of the console applications. Please review it a bit closer, and consider the notes I left.
zul: would you like to join app-backup?
Robin: Do you know what's up with zul? Last seen on irc +/- two weeks, the bacula bugs are getting really old and the interest doesn't seem to be low. I tried to find a developer via irc who is using bacula - no luck. Chris White doesn't use it, but showed interest.
I dont have interest anymore bacula so i you want its yours. I dont use it at all either, and I dont have time for it. Sorry chuck
Carsten: I'll grab beejay for it, since he noted in his blog that he uses it on his home network.
only want to say that the version 1.36.3 was released in meantime
*** Bug 93789 has been marked as a duplicate of this bug. ***
Created attachment 59723 [details] bacula-1.36.3.ebuild working ebuild for 1.36.3 with bacula-clientonly, bacula-console, and bacula-split USE flags, as well as support for environment variables to specify db user, password, and host when checking db version in pkg_postinst()
Created attachment 59724 [details] files/digest-bacula-1.36.3
beejay: interested in maintaining bacula?
Created attachment 60836 [details] updated 1.36.3 ebuild Reworked use flags some more, cleaning up alot of 'if use' in the configuration. Added logrotate and logwatch local flags to optionally suppress unwanted installation of respective scripts. Shortened the verbosity of alot of einfo, particularly in pkg_postinst(). Replaced a few 'make's with 'emake's.
Created attachment 60838 [details] updated 1.36.3 ebuild (for init-depends patch) patched init scripts which previously omitted depend()
Created attachment 60839 [details, diff] init scripts depend() patch package-included split init scripts lacked depend() and caused errors on startup if started before net or db
Created attachment 60880 [details] another 1.36.3 ebuild corrected bug in attachment #60838 [details] ebuild around lines 290-295 where db checks assumed that at least one db was set in use -- added an else to force mysql if no db specified in use.
Created attachment 60916 [details, diff] patch to add working Makefile for cd rescue package This lets make handle building the static file daemon and installing the rescue package directory instead of having so many manual commands in the ebuild.
Created attachment 60917 [details] yet another 1.36.3 ebuild uses attachment #60916 [details, diff] patch and trims down alot of manual cd/cp/mkdir/etc, letting (e)make do these things instead of them (redundantly) taking up space in the ebuild.
Created attachment 60926 [details] fixed 1.36.3 ebuild My last couple of ebuild submissions had a bug in the detection of db use flags which failed to properly update ${myconf} when only one db was in use. This is now fixed around lines 115-131.
Created attachment 60930 [details] fixed 1.36.3 ebuild fixed syntax errors in postinst
Created attachment 61577 [details, diff] fixed rescue-makefile patch previously submitted bacula-1.36.3-rescue-makefile.patch caused make to fail during 'make install' and is now fixed
Me and hadfield have just commited a new 1.36.3-r1 ebuild. Please test and report any bugs.
Ciaranm informed me that any combination of USE flags - conflicting, unspecified, whatever - should not cause emerge to die. In your ebuild you have removed some DEPEND and USE checks that were put there for that very reason. Either what ciaranm told me was wrong, or your ebuild deviates from policy. Why do you have so much manual 'sed ...' in the ebuild? Why have you removed the split init script feature requested by users?
