New: [rhinstaller/blivet/pulls/227 master] warn on failure to scan partitions
by dashea
I had a Fedora install fail due to https://bugzilla.redhat.com/1263832,
and I wonder if we should be treating failures at this point in the code
as warnings rather than errors. This would permit blivet to continue
scanning other devices.
This would seem to be in line with the comments at the beginning of this
section of code, which suggest that the desired behavior is to ignore
the device rather than throwing an exception:
# Ignore partitions on:
# - devices we do not support partitioning of, like logical volumes
# - devices that do not have a usable disklabel
# - devices that contain disklabels made by isohybrid
--
To view this pull request on github, visit https://github.com/rhinstaller/blivet/pull/227
8 years, 8 months
[PATCH 1/2] Run substitution checks on the right kickstart file.
by Chris Lumens
${ks} is the input kickstart file, which is the .ks.in version.
${ksfile} is the one after prepare has been run, which might involve
doing other modifications first. This only affects those few kickstart
tests that do something complex in their prepare function.
---
tests/kickstart_tests/scripts/run_one_ks.sh | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/kickstart_tests/scripts/run_one_ks.sh b/tests/kickstart_tests/scripts/run_one_ks.sh
index f0b43aa..9e86dfa 100755
--- a/tests/kickstart_tests/scripts/run_one_ks.sh
+++ b/tests/kickstart_tests/scripts/run_one_ks.sh
@@ -80,7 +80,7 @@ runone() {
# Check that the prepared kickstart is free of substitution markers. Normally
# the substitutions are run by run_kickstart_tests.sh, but prepare has a chance
# to run them too. If both of those left any @STUFF@ strings behind, fail.
- unmatched="$(grep -o '@[^[:space:]]\+@' ${ks} | head -1)"
+ unmatched="$(grep -o '@[^[:space:]]\+@' ${ksfile} | head -1)"
if [ -n "$unmatched" ]; then
echo "RESULT:${name}:FAILED:Unsubstituted pattern ${unmatched}"
cleanup ${tmpdir}
--
2.4.3
8 years, 8 months
New: [rhinstaller/blivet/pulls/226 master] lvactivate when lvmetad is active
by dashea
When lvmetad is running and the last PV in a VG is activated, an asynchronous system unit (`lvm2-pvscan@...service`) is scheduled to update LVM's view of things and to auto-activate the LVs. During the interim between that unit being scheduled and run, LVM behaves as if the PV were still inactive. So, when we're deactivating/activating the PVs we should wait for auto-activation instead of plowing ahead and trying to activate the LVs ourselves. This has caused several bugs in which we activate an MD PV, then try to activate an LV, and get an error from LVM saying the VG doesn't exist.
--
To view this pull request on github, visit https://github.com/rhinstaller/blivet/pull/226
8 years, 8 months
[anaconda][rhel7-branch/master/f23-branch][PATCH 0/2] Environment selection fixes
by Martin Kolman
The first patch contains the main fix - fixing the issue with environment
reverting to "minimal install" after reentering the software spoke during
manual installation.
The second patch fixes a related issue I have found when testing the first patch,
where the software spoke status would be incorrect after an environment has been selected
manually but there is no %packages section in kickstart.
Martin Kolman (2):
Keep environment selection when reentering the software spoke
(#1261393)
Show correct status for no %packages with selected environment
(#1261393)
pyanaconda/ui/gui/spokes/software.py | 29 +++++++++++++++++------------
1 file changed, 17 insertions(+), 12 deletions(-)
--
2.4.3
8 years, 8 months