summaryrefslogtreecommitdiffstats
path: root/data/systemd
Commit message (Collapse)AuthorAgeFilesLines
* remove anaconda-cleanup-initramfs.serviceWill Woods2013-02-114-51/+1
| | | | | | | | | | This service operated under the assumption that the initramfs would be saved at /run/initramfs. Otherwise, it doesn't do anything. And initramfs doesn't get saved to /run/initramfs anymore. So this service no longer does anything. This patch removes it.
* Fixes for PkgWrangler review.David Cantrell2012-11-301-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Specifically: anaconda.src:183: W: unversioned-explicit-obsoletes booty anaconda.src:262: E: hardcoded-library-path in /lib/systemd/system/* anaconda.src:263: E: hardcoded-library-path in /lib/systemd/system-generators/* anaconda.src:287: E: files-attr-not-set anaconda.src:288: E: files-attr-not-set anaconda.src:289: E: files-attr-not-set anaconda.src:290: E: files-attr-not-set anaconda.src:293: E: files-attr-not-set anaconda.src:294: E: files-attr-not-set anaconda.src:295: E: files-attr-not-set anaconda.src:296: E: files-attr-not-set anaconda.src:299: E: hardcoded-library-path in /usr/lib/dracut/modules.d/80%{name} anaconda.src:299: E: files-attr-not-set anaconda.src:300: E: hardcoded-library-path in /usr/lib/dracut/modules.d/80%{name}/* anaconda.src:300: E: files-attr-not-set anaconda.src:624: W: macro-in-%changelog %packages anaconda.src:1234: W: macro-in-%changelog %include anaconda.src:1680: W: macro-in-%changelog %{nil} anaconda.src:1733: W: macro-in-%changelog %post anaconda.src:2160: W: macro-in-%changelog %{nil} anaconda.src:2325: W: macro-in-%changelog %includes anaconda.src:2564: W: macro-in-%changelog %packages anaconda.src:2565: W: macro-in-%changelog %post anaconda.src:2592: W: macro-in-%changelog %pre anaconda.src:2788: W: macro-in-%changelog %pre Updated data/ and data/systemd/ to install to /usr/lib rather than /lib since the UsrMove thing happened. Removed the newt and slang BuildRequires since we do not use those anymore.
* Add a service to run anaconda directly on the ttyJesse Keating2012-10-103-2/+26
| | | | | | | | | When installing on s390x, we just want to run pure anaconda on the tty. This will either detect s390x and print out ssh details for the user to ssh in to continue the install (and get tmux), or if a kickstart was provided and the RUNKS option was used it will just blast through the kickstart in cmdline mode. Tmux doesn't work in x3270, the terminal for s390x, so we have to avoid using it.
* Use the real path to dracut-lib.sh (#851362)Jesse Keating2012-09-061-1/+1
| | | | | | This goes with a lorax patch that keeps the file around. Since dracut doesn't expose these files via /run/ anymore we have to keep around the file(s) we want to use from stage2 environment.
* Correct the location of the tmux config file.Chris Lumens2012-08-031-1/+1
| | | | | | Otherwise, it won't start and thus anaconda won't start. The other way to fix this would be to add stuff to lorax to move the file into place, but I don't want to make lorax changes.
* Run anaconda inside tmuxWill Woods2012-08-026-11/+22
| | | | | | | | | | | | | | | | | | | | | | | | Having a shell on tty2 or hvc1 is convenient.. if you have tty2 or hvc1. But not everyone does - think s390, or POWER, or remote virt guests. Instead, let's run anaconda inside a tmux session, with windows pre-configured with a shell, anaconda.log, program.log, and storage.log. We set this up by using two services: anaconda.service, and anaconda-tmux@.service. anaconda.service starts anaconda (and starts watching the logs) inside a detached tmux session, on screen 1. It also starts a shell on screen 2, and 'tail -F' on anaconda.log, storage.log, and program.log on screens 3, 4, and 5. anaconda-tmux@.service will be instantiated on the default console (tty1, hvc0, ttyS0, whatever) and connect to the running tmux session, which will show screen 1 - and thus the text UI (if any) appears. Exiting tmux just restarts anaconda-tmux@.service, which reconnects to the existing session, so you can't accidentally kill the installer by leaving tmux.
* Re-remove fedora-import-stateWill Woods2012-08-022-27/+0
| | | | | Commit c8c14c9 deleted these two files, but they got resurrected in some merge somehow. Delete them again, since we don't need them.
* Actually start anaconda-cleanup-initramfs.serviceWill Woods2012-06-211-0/+1
| | | | | | | Rule #1 of Computing: It Works Better When You Turn It On. Make anaconda.target request anaconda-cleanup-initramfs.service so it actually gets started.
* compress /run/initramfs/usr in initramfs cleanupWill Woods2012-06-192-0/+18
| | | | | | | | | We don't actually need the contents of /run/initramfs/usr while inside anaconda, so we can compress it and stick it in /boot, saving ~28MB RAM. To be able to shut down properly, we'll need dracut-shutdown.service (and its dependencies) in the runtime; there should be an associated lorax patch that accomplishes this.
* Add anaconda-cleanup-initramfs.serviceWill Woods2012-06-193-1/+39
| | | | | | | | | This adds anaconda-cleanup-initramfs.service, so we can clean unnecessary/redundant files out of /run/initramfs to save RAM. The initial cleanup strategy is to remove kernel modules, firmware, and the ssl certificate store, because the anaconda runtime has all these things already. This saves about 40MB RAM.
* Drop refs to dev-%i.device from anaconda-shell@.service (#826984)Will Woods2012-06-131-2/+1
| | | | | Apparently systemd/udev no longer create units for tty* devices; without those units anaconda-shell@.service wouldn't start.
* Move sshpw handling out of AnacondaJesse Keating2012-06-081-0/+1
| | | | | | | | This creates an ExecStartPre on the anaconda-sshd.service that will check for a kickstart entry for sshpw. It will add/modify users accordingly prior to launching the sshd service. Since now sshpw and sshd bring up happens outside of and before anaconda starts, we can remove sshd.py and any reference to it.
* Use systemd to startup sshd when desiredJesse Keating2012-06-083-1/+23
| | | | | | | This adds another chunk to the anaconda-generator that will figure out what services go where. In this case, we enable the sshd service when inst.sshd or sshd is found as a command line option, or if we're on s390x (because the console sucks there).
* Set the python path within anacondaJesse Keating2012-05-301-1/+1
| | | | | | | | Instead of manipulating the path outside of anaconda, which can get tricky when starting anaconda from ssh, instead update the path inside anaconda itself. This means we can't get things like os, sys, etc.. from /tmp/updates/, but an updates image can replace these modules by using the full path to them (e.g. /usr/lib/python...)
* Merge 'f17-branch'Will Woods2012-03-305-14/+52
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This brings in the 'noloader' changes from f17-branch. The changelogs have been dropped from the specfile, but that was somewhat intentional. Conflicts: anaconda anaconda.spec.in configure.ac data/linuxrc.s390 data/systemd/Makefile.am data/systemd/anaconda.service data/systemd/anaconda.target data/systemd/anaconda@.service data/systemd/loader.service dracut/Makefile.am dracut/anaconda-lib.sh dracut/anaconda-netroot.sh dracut/kickstart-genrules.sh dracut/module-setup.sh dracut/parse-anaconda-kickstart.sh dracut/parse-anaconda-options.sh dracut/parse-anaconda-repo.sh dracut/parse-kickstart dracut/repo-genrules.sh po/anaconda.pot
| * Let systemd handle terminal setup, fix possible race with NMWill Woods2012-03-215-12/+52
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Add a systemd-generator script that puts anaconda on the correct console tty, then puts shells on tty2 and the first virtualization console (unless we're using it for anaconda). Adding "TTYPath" to the systemd services means systemd takes care of setting up all the terminal stuff - picking the right TERM value, setting the keyboard mode to K_UNICODE, and all that esoteric weirdness. Also, move anaconda's dependencies into anaconda.target, and make anaconda start *after* anaconda.target, just so we can be sure that e.g. NetworkManager is active before anaconda starts. (as a bonus, it also shuts off plymouth correctly, just in case we ever want to have a anaconda-specific bootsplash)
| * anaconda.service Wants=NetworkManager.serviceWill Woods2012-03-161-2/+2
| | | | | | | | | | | | | | | | If we brought up any interfaces in stage1, we need NetworkManager to take them over before anaconda starts. We *could* do this conditionally and delay NM startup until needed, but this is the simplest fix for now.
| * fedora-import-state.service is in initscripts nowWill Woods2012-03-163-30/+1
| | | | | | | | | | initscripts-9.35-1 contains fedora-import-state.service, so we can drop this from here.
| * a couple small cleanups/fixes for fedora-import-state.serviceWill Woods2012-03-162-2/+7
| |
| * anaconda-shell service tweaksWill Woods2012-03-161-1/+1
| | | | | | | | | | following getty@.service, make sure we run the shell after plymouth quits, etc.
| * add fedora-import-state.service (fix NFS root: #799989)Will Woods2012-03-164-2/+26
| | | | | | | | | | | | | | | | | | | | | | dracut writes out an ifcfg-ethX file and a dhclient lease file for the interface used to mount the runtime. We need to pass that along to NetworkManager so it won't bring down the interface we're using for our root device. This service copies the contents of the /run/initramfs/state directory into its intended place, as would normally be done in fedora-readonly for NFS readonly-root systems.
| * Move from loader.service to anaconda.serviceWill Woods2012-03-164-5/+5
| | | | | | | | | | Replace loader.service with anaconda.service, which starts anaconda directly. Make necessary changes to related files.
* | Merge branch 'noloader'Will Woods2012-03-305-6/+6
|\ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Merge the noloader branch into master. Hooray! Conflicts: data/systemd/Makefile.am data/systemd/anaconda.target data/systemd/fedora-import-state loader/loader.c loader/net.c
| * | a couple small cleanups/fixes for fedora-import-state.serviceWill Woods2012-03-082-2/+7
| | |
| * | anaconda-shell service tweaksWill Woods2012-03-071-1/+1
| | | | | | | | | | | | | | | following getty@.service, make sure we run the shell after plymouth quits, etc.
| * | add fedora-import-state.service (fix NFS root: #799989)Will Woods2012-03-074-2/+26
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | dracut writes out an ifcfg-ethX file and a dhclient lease file for the interface used to mount the runtime. We need to pass that along to NetworkManager so it won't bring down the interface we're using for our root device. This service copies the contents of the /run/initramfs/state directory into its intended place, as would normally be done in fedora-readonly for NFS readonly-root systems.
| * | Move from loader.service to anaconda.serviceWill Woods2012-02-154-5/+5
| |/ | | | | | | | | Replace loader.service with anaconda.service, which starts anaconda directly. Make necessary changes to related files.
* / Transfer network state from dracut to anaconda to keep NFS workingWill Woods2012-03-124-2/+31
|/ | | | | | | We need to prevent NM to shut down interfaces, because we have NFS mounted squash.img in NFS based install. Resolves: rhbz#799989
* it is anaconda-shell@.service or we can't build.Ales Kozumplik2011-09-231-1/+1
| | | | see also 6239b2ddc9a595059d20f4672e0140391382a860.
* Remove tmp.mount (systemd handles this for us now)Will Woods2011-09-212-10/+1
|
* Move dependency info into the unit filesWill Woods2011-09-213-12/+4
|
* move anaconda-shell.service to the correct filenameWill Woods2011-09-211-0/+0
|
* make anaconda-shell.service a template, put it on tty2 & hvc1Will Woods2011-09-212-5/+8
| | | | | | We're also passing "-l /bin/bash -o '--login'", which does the no-login thing for us, so we can remove the silly hack where we replace /bin/login with a shell script that runs bash.
* systemd: anaconda.target wants rsyslog.serviceAles Kozumplik2011-09-081-0/+1
| | | | | | Else systemd doesn't start rsyslogd and all the logs get dumped to ttys. Resolves: rhbz#736125
* ConditionKernelCommandLine is a setting for Unit, not Service.Chris Lumens2011-08-101-1/+1
|
* simplify anaconda.target/loader.service requirementsWill Woods2011-08-093-5/+6
| | | | | | | | | A lot of the stuff loader.service currently requires (udev, syslog, dbus, etc.) is started by basic.target, so let's have anaconda.target start after basic.target (much like multi-user.target) and simplify loader.service's requirements. (also, there's no such thing as dbus.target)
* make anaconda-shell.service more like getty@.serviceWill Woods2011-08-091-0/+6
|
* Remove KillMode= from systemd control files.Chris Lumens2011-06-282-2/+0
| | | | | process-group was an old setting that's no longer valid, and the default setting is fine for us.
* Add a python program to record memory usage during installation.Chris Lumens2011-05-163-3/+11
| | | | | | This program only runs during installation if the "debug" command line option is provided. It writes out a /tmp/memory.dat file which can be copied off the system and processed with the provided gnuplot script to display a graph.
* Remove from init.c/loader.c things that systemd does for us.Chris Lumens2011-04-123-5/+14
| | | | | | | * We don't need to populate /dev. * We don't need to mount filesystems. * We don't need to start dbus or udevd. * We don't need to start a shell in multiple places.
* Add the unit files necessary to have systemd start loader.Chris Lumens2011-04-124-0/+59