Fixes problems found by set -eu and pipefail, including:
-Many unset variables
-Commands that can fail under normal circumstances, which breaks
with set -e. This change swallows those expected errors to allow
our existing error code to handle them.
-The dkms element was not finding Fedora kernel versions correctly.
This may be an issue for other distros too, but since Fedora was
working fine without this functionality I only changed it to print
a warning message rather than failing the build when it happens.
-The ramdisk init script will not be set -eu because if it fails
the result is a kernel panic, which can be tricky to debug.
However, in testing with set -e a few failing commands were found
and have been fixed in this patch.
Change-Id: I44cf98dfc80cfcaec54b88cc83be80a3dbf2cec3
The version of depmod in busybox does not work if two versions of
the same kernel version are available, as search paths are not
honoured correctly. However, we don't need to actually call
depmod as it's done during image building anyway.
Change-Id: Idd3cd5854e7c86e7ca1f2b0e50758d7db79ab6e7
Co-Author: matthew.gilliard@hp.com
Closes-bug: #1302856
tgtd returns execution control and backgrounds itself almost immediately
and before it has made it's listening socket available. This can cause a
race condition as the tgtd socket is not available when tgtadm is run,
resulting in an error:
failed to send request hdr to tgt daemon
Add a function to check if the socket is available before moving on to
calling tgtadm, and a wait_for helper function we can use. We'll
check for the socket every 0.5 seconds, for up to 5 seconds.
I'm seeing this issue on almost every deploy using a ramdisk built from
Fedora 20. I'm not sure if something has changed in tgtd, but this
behavior is documented since Fedora 18 at least. In the systemd script
for tgtd, there is actually "sleep 5" to work around the problem.
See Also: https://bugzilla.redhat.com/show_bug.cgi?id=848942
Change-Id: Iffa9fc63393309ca653d592dff17316ecbea3e09
We can no longer rely on the name of the udevd binary as a hint for its
version, so instead we query the binary and behave accordingly.
Change-Id: Id4c869dfeeca5ed970169343b6b5f0f2c4499ca3
We can now actually override the $TROUBLESHOOT environment variable in
ramdisks, by not declaring it as readonly. Yay for insufficient
end-to-end testing.
Change-Id: I026c9dbdd411bf3eb945d5b18188eb1eedf3850a
We now allow operators to drop into a troubleshooting shell while
booting a deploy ramdisk, without having to have pre-configured their
desire to do so ahead of time, by means of a 10 second pause for them to
press a key.
Change-Id: I05ffa1ebaa95c83dee7bd2a2f52ba4c08928bb10
We now run the network bringup portion of the ramdisk init from an
init.d script, thus allowing image builders to inject code into the init
script before that point (e.g. configure network hardware).
The environment variables used by the network bringup code are retained
in the base init script so they can be overridden by init.d fragments.
Change-Id: I1ef0bb21e7f26c0ff3f02266f853ce5402bcb94d
Closes-Bug: #1252023
Ramdisks are now built inside a chroot which is built by the normal
image build process. Doing so improves our independence of the
precise state of the build host.
This fixes bug 1194055.
Change-Id: Ibc254fbb9e7b404b5f38c1b35bcde8a4136e8e28
