summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* Fix default definitions of some payload class methods.Chris Lumens2012-08-281-1/+7
| | | | | | Various parts of anaconda either call these methods or expect them to return certain values (so, not NotImplementedError) and it's just easier to have them return default values than create new implementations.
* Add a spaceRequired property for LiveImagePayload.Chris Lumens2012-08-281-0/+5
|
* getDirSize should stay on a single filesystem, not look at submounts.Chris Lumens2012-08-281-0/+2
|
* Don't look for existing installations on live devices.Chris Lumens2012-08-281-1/+2
| | | | Or really, any devices we can't actually use.
* We don't need image_file in the live payload.Chris Lumens2012-08-282-8/+2
| | | | | / is always mounted for us anyway, and since we're using rsync it's just a simple matter of copying from that directory to our destination.
* Now that we're using rsync, the livecd and rootfs do not have to match.Chris Lumens2012-08-282-23/+0
|
* Disable software selection and source spokes on live installs.Chris Lumens2012-08-282-0/+9
|
* Fix args to LiveImagePayload.setup (#852272).Chris Lumens2012-08-281-1/+1
|
* require anaconda-widgetsBrian C. Lane2012-08-281-0/+1
|
* Handle already mounted optical devices (#851274)Jesse Keating2012-08-271-6/+11
| | | | | | | This prevents trying to mount the device a second time. If it's already mounted, and is a valid source of repodata, use it from the path it's already mounted on. Otherwise go ahead and mount the selected optical drive where we want it.
* Return full device object of selected optical driveJesse Keating2012-08-271-2/+3
| | | | | The full device is more useful than just the path. Callers can get the path if they want it.
* Add a method to determine if device is mountedJesse Keating2012-08-271-0/+15
| | | | This will return a list of places a particular device might be mounted.
* anaconda-cleanup: fix DeviceTree argsBrian C. Lane2012-08-271-1/+1
|
* Unset install_device if repo setup failsJesse Keating2012-08-271-0/+1
| | | | | We're unsetting method there too, we need to unset the install_device so that we don't try and mount it later.
* _peopleRepositoriesFilter -> _peopleRepositoriesFilterEntry (#852182).Chris Lumens2012-08-271-1/+1
|
* on_*_changed callbacks take one argument, not two.Chris Lumens2012-08-271-2/+2
|
* Use the correct icon size constant.Chris Lumens2012-08-271-4/+4
|
* remove dead code (setMethodstr, expandFTPMethod)Will Woods2012-08-272-38/+1
| | | | | | Nothing uses a method/repo of 'cdrom://<dev>:<path>' anymore, and nothing passes a method like '@/tmp/filename', so this code is all useless.
* parse-kickstart: update some TODO commentsWill Woods2012-08-271-6/+1
| | | | | A bunch of these comments don't apply anymore - update them so they reflect reality
* parse-kickstart: simplify loggingWill Woods2012-08-271-22/+3
| | | | Systemd handles the tricky bits of logging for us now. Yay!
* enable fastestmirror yum plugin (#849797)Brian C. Lane2012-08-271-1/+1
|
* networking: remove Network() objectRadek Vykydal2012-08-273-46/+47
| | | | | | - We can add dhcp activation in networkInitialize in thread later - Also fixing merge (newui->master) of https://lists.fedorahosted.org/pipermail/anaconda-patches/2012-June/000023.html
* networking: use ksdata.network.hostname instead of actual installer hostnameRadek Vykydal2012-08-273-16/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | ksdata.network.hostname which is new property (there is a related kickstart patch) obtaining the hostname from network commands (ksdata.network.network list of NetworkData objects). The goal is to use this data instead of getting actual hostname of installer (getHostname function) notes: - ksdata.network.hostname value will be initalized in networkInitialize (following patch) - getHostname (actual hostname of system) is still used in dracutSetupArgs, we need to get ksdata passed somehow from bootloader.py:set_boot_args() - Hostname of installer environment is still set in kickstart. We used to do it because of mdadm member naming (640743) but this is not needed anymore - mdadm actually doesn't use it. If it starts to again, or does it in some special conditions, we can pass ksdata.network.hostname (as we do for default vg names) using --homehost option. We might need to set the hostname for some installed package though, so I am keeping it. - We should think about where to set hostname in GUI. This should happen before writing storage (needed for vg, mdmember names), or we can drop default naming based on hostname.
* networking: consolidate writing/copying of configuration filesRadek Vykydal2012-08-271-42/+22
|
* networking: 70-persistent-net.rules doesn't exist anymore.Radek Vykydal2012-08-271-4/+0
| | | | See udev's 828136.
* networking: disable ipv6 directly in installed system config fileRadek Vykydal2012-08-271-9/+6
| | | | Instead of creating the file in installer and copying it to system.
* networking: mirror end-of-installation network config tweaks in ksdata.Radek Vykydal2012-08-273-5/+16
| | | | | | | ksdata are updated in NetworkSpoke.apply from ifcfg files. At the end of installation we need to modify ifcfg files (network configuration) for devices used by storage and to set ONBOOT defaults. So update also ksdata.
* networking: write configuration in doInstallRadek Vykydal2012-08-272-0/+14
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Short story: Ifcfg files are (still) copied from installer environment instead of generating from ksdata[1]. Details: Ifcfg files can be created during installation - in dracut: based on boot options and kickstart - in anaconda: missing default ifcfg files[2] - in GUI using nm-c-e Going through ksdata here (update ksdata in spoke.apply and write ifcfg files from ksdata at the end of install - writeNetworkConf in doInstall [3]) - which is what we would generally do for other spokes - is not necessary, and has some caveats: We'd need to update ks network command to mirror all configuration doable in ifcfg files with nm-c-e (plus wireless credentials stored in keys-<device> files, routing stored in yet another network-scripts files) and keep up with every update of nm-c-e options instead of letting nm-c-e just write the configuration for us. Also NM can update ifcfg files on its own after they are written - another opprotunity for missync. [1] Beside ifcfg files we also copy /etc/sysconfig/network-scripts/keys-<device> files. We might want to copy also routing files created in nm-c-e. [2] this should go to dracut IMO. [3] such would be the chain: - dracut writes ifcfg files - NM can update ifcfg when activating device (does anaconda need to update ksdata ?) - anaconda can write missing default ifcfg files - nm-c-e can modify ifcfg files (GUI configuration) - anaconda stores ifcfg to ksdata (spoke.apply) - anaconda dumps ksdata to ifcfg on /mnt/sysimage (doInstall) - anaconda modifies configuration for storage devices, and to set ONBOOT default - currently this can't precede the previous step because we are changing NM_CONTROLLED which is not mirrored in ksdata, so we need also to sync ksdata in this step
* Add mounts before swaps so the default selection is a mount.David Lehman2012-08-241-6/+6
|
* Use MB if a new mountpoint size does not include a unit spec. (#850839)David Lehman2012-08-241-1/+7
|
* Correctly handle partitions with sizes smaller than 500MB. (#850839)David Lehman2012-08-241-4/+13
|
* Don't include removed devices in Storage.unusedDevices.David Lehman2012-08-241-0/+3
|
* Handle SameSizeSet growth trimming when all members are too large.David Lehman2012-08-241-2/+7
|
* Add several missing yum lock aqcuisitions. (#851212)David Lehman2012-08-241-12/+17
| | | | | Do locking inside release since it gets called from outside of YumPayload.
* Offer completions for new mountpoints.David Lehman2012-08-242-2/+25
|
* Add old_source checking for closest mirror and url methods too (#851336).Chris Lumens2012-08-241-0/+6
| | | | | | This way if the user starts with one of those methods and doesn't change anything before hitting back, we won't waste time trying to redownload everything.
* Revert "Only use mounted media that has repodata"Jesse Keating2012-08-241-5/+4
| | | | | This reverts commit d82ee4a818be9db7b65282fad7de3c8bf872a6f3. There is a better way to do this.
* Only use mounted media that has repodataJesse Keating2012-08-241-4/+5
| | | | | | | | This recovers some logic from yuminstall.py, only use /run/install/repo if there is metadata there. This is related to bug 851274 , but only fixes part of it. It gets booting from boot.iso working again, but leaves booting from DVD broken.
* _bootloaderClass -> bootloaderClass for some platforms (#848173).Chris Lumens2012-08-231-4/+4
|
* Make the storage info bar clickable to reveal error messages.Chris Lumens2012-08-232-30/+28
|
* Move the software-specific error message out of the DetailedErrorDialog class.Chris Lumens2012-08-233-5/+12
|
* Add a gui password spokeJesse Keating2012-08-233-0/+320
| | | | | | | | | This spoke sets the root password, or if left blank disables the root login. Some level of validation is done and the user is trapped until they get it right or press back twice on a weak password. capslock monitoring is not enabled currently, but there is some stub code commented out that deals with it.
* Put traceback reports on a diet.Chris Lumens2012-08-231-22/+7
| | | | | | | | | This updates what attributes get written out to the traceback file for the newui. In particular, it avoids writing out all the yum stuff that makes the traceback file grow over 2 MB and take nearly a minute to pickle. It's entirely possible this is either skipping too much or writing out too much still, but we can work on that over time.
* New version.Chris Lumens2012-08-223-87/+169
|
* Do another _main_window -> main_window change.Chris Lumens2012-08-221-1/+1
|
* Mark the storage category title for translation.Chris Lumens2012-08-221-1/+3
|
* _actions should be set up in the __init__ method.Chris Lumens2012-08-221-1/+1
| | | | This prevents a traceback from happening when handling tracebacks.
* Don't require hfs-tools on RHEL (#849987).Chris Lumens2012-08-221-0/+2
|
* dracut: remove workarounds for broken splitsep()Will Woods2012-08-222-5/+2
| | | | | | | | | | the splitsep function was buggy before, but it was fixed in dracut 019 so we don't need these workarounds anymore. See upstream dracut commit f7cadaa843498c4b986f8a030fab39002ad108b6. Related: 844c0c156ed2f27920257d86690f3017727f398e Related: 700e985082b03abcbd3009de81885b8fbf72818b
* dracut: update Requires: in specWill Woods2012-08-221-1/+1
| | | | | | | | | We updated the anaconda dracut module to work with "newer" dracut; all the features/changes that would actually cause compatibility problems are present in dracut 019 and later. (The systemd changes are in dracut 020 but existing anaconda-dracut code is still backward-compatible with 019)
generated by cgit (git 2.34.1) at 2025-09-28 01:48:03 +0000