rpm-ostree/tests/common/libvm.sh
Jonathan Lebon c7a9c3b1dd Rework vmcheck to use kola spawn, move off of PAPR
There's a lot going on here, but essentially:

1. We change the `vmcheck` model so that it always operates on an
   immutable base image. It takes that image and dynamically launches a
   separate VM for each test using `kola spawn`. This means we can drop
   a lot of hacks around re-using the same VMs.
2. Following from 1., `vmoverlay` now takes as input a base image,
   overlays the built rpm-ostree bits, then creates a new base image. Of
   course, we don't have to do this in CI, because we build FCOS with
   the freshly built RPMs (so it uses `SKIP_VMOVERLAY=1`). `vmoverlay`
   then will be more for the developer case where one doesn't want to
   iterate via `cosa build` to test rpm-ostree changes. I say "will"
   because the functionality doesn't exist yet; I'd like to enhance
   `cosa dev-overlay` to do this. (Note `vmsync` should still works just
   as before too.)
3. `vmcheck` can be run without building the tree first, as
   `tests/vmcheck.sh`. The `make vmcheck` target still exists though for
   finger compatibility and better meshing with `vmoverlay` in the
   developer case.

What's really nice about using kola spawn is that it takes care of a lot
of things for us, such as the qemu command, journal and console
gathering, and SSH.

Similarly to the compose testsuites, we're using parallel here to run
multiple vmcheck tests at once. (On developer laptops, we cap
parallelism at `$(nproc) - 1`).
2019-12-13 19:18:30 +01:00

708 lines
20 KiB
Bash

# Source library for installed virtualized shell script tests
#
# Copyright (C) 2016 Jonathan Lebon <jlebon@redhat.com>
#
# This library is free software; you can redistribute it and/or
# modify it under the terms of the GNU Lesser General Public
# License as published by the Free Software Foundation; either
# version 2 of the License, or (at your option) any later version.
#
# This library is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
# Lesser General Public License for more details.
#
# You should have received a copy of the GNU Lesser General Public
# License along with this library; if not, write to the
# Free Software Foundation, Inc., 59 Temple Place - Suite 330,
# Boston, MA 02111-1307, USA.
# prepares the VM and library for action
vm_setup() {
export VM=${VM:-vmcheck}
export SSH_CONFIG=${SSH_CONFIG:-${topsrcdir}/ssh-config}
SSHOPTS="-o User=root -o ControlMaster=auto \
-o ControlPath=/dev/shm/ssh-$VM-$(date +%s%N).sock \
-o ControlPersist=yes"
# If we're provided with an ssh-config, make sure we tell
# ssh to pick it up.
if [ -f "${SSH_CONFIG}" ]; then
SSHOPTS="${SSHOPTS} -F ${SSH_CONFIG}"
fi
export SSHOPTS
export SSH="ssh ${SSHOPTS} $VM"
export SCP="scp ${SSHOPTS}"
}
# prepares a fresh VM for action via `kola spawn`
vm_kola_spawn() {
local outputdir=$1; shift
exec 4> info.json
mkdir kola-ssh
setpriv --pdeathsig SIGKILL -- \
env MANTLE_SSH_DIR="$PWD/kola-ssh" kola spawn -p qemu-unpriv \
--qemu-image "${topsrcdir}/tests/vmcheck/image.qcow2" -v --idle \
--json-info-fd 4 --output-dir "$outputdir" &
# hack; need cleaner API for async kola spawn
while [ ! -s info.json ]; do sleep 1; done
local ssh_ip_port ssh_ip ssh_port
ssh_ip_port=$(jq -r .public_ip info.json)
ssh_ip=${ssh_ip_port%:*}
ssh_port=${ssh_ip_port#*:}
cat > ssh-config <<EOF
Host vmcheck
HostName ${ssh_ip}
Port ${ssh_port}
StrictHostKeyChecking no
UserKnownHostsFile /dev/null
EOF
SSH_CONFIG=$PWD/ssh-config
# XXX: should just have kola output the path to the socket
SSH_AUTH_SOCK=$(ls kola-ssh/agent.*)
export SSH_CONFIG SSH_AUTH_SOCK
# Hack around kola's Ignition config only setting up the core user; but we
# want to be able to ssh directly as root. We still want all the other goodies
# that kola injects in its Ignition config though, so we don't want to
# override it. `cosa run`'s merge semantics would do nicely.
ssh -o User=core -F "${SSH_CONFIG}" vmcheck 'sudo cp -RT {/home/core,/root}/.ssh'
vm_setup
# XXX: hack around https://github.com/systemd/systemd/issues/14328
vm_cmd systemctl mask --now systemd-logind
# Some tests expect the ref to be on `vmcheck`. We should drop that
# requirement, but for now let's just mangle the origin
local deployment_root
vm_cmd ostree refs --create vmcheck "$(vm_get_booted_csum)"
deployment_root=$(vm_get_deployment_root 0)
vm_cmd "sed -ie '/^refspec=/ s/=.*/=vmcheck/' ${deployment_root}.origin"
vm_cmd "sed -ie '/^baserefspec=/ s/=.*/=vmcheck/' ${deployment_root}.origin"
vm_cmd systemctl try-restart rpm-ostreed
# also move the default yum repos, we don't want em
vm_cmd mv /etc/yum.repos.d{,.bak}
vm_cmd mkdir /etc/yum.repos.d
}
# $1 - file to send
# $2 - destination path
vm_send() {
$SCP ${1} ${VM}:${2}
}
# $1 - destination path
vm_send_inline() {
f=$(mktemp -p $PWD)
cat > ${f}
vm_send ${f} ${1}
rm -f ${f}
}
vm_shell_inline() {
script=$(mktemp -p $PWD)
echo "set -xeuo pipefail" > ${script}
cat >> ${script}
vm_send ${script} /tmp/$(basename ${script})
rm -f ${script}
vm_cmd bash /tmp/$(basename ${script})
}
# rsync wrapper that sets up authentication
vm_raw_rsync() {
local rsyncopts="ssh -o User=root"
if [ -f "${SSH_CONFIG}" ]; then
rsyncopts="$rsyncopts -F '${SSH_CONFIG}'"
fi
rsync -az --no-owner --no-group -e "$rsyncopts" "$@"
}
vm_rsync() {
if ! test -f .vagrant/using_sshfs; then
pushd ${topsrcdir}
vm_raw_rsync --delete --exclude target/ --exclude bindgen-target/ --exclude .git/ . $VM:/var/roothome/sync
popd
fi
}
# run command in vm as user
# - $1 username
# - $@ command to run
vm_cmd_as() {
local user=$1; shift
# don't reuse root's ControlPath
local sshopts="-o User=$user"
if [ -f "${SSH_CONFIG}" ]; then
sshopts="$sshopts -F ${SSH_CONFIG}"
fi
ssh $sshopts $VM "$@"
}
# run command in vm
# - $@ command to run
vm_cmd() {
$SSH "$@"
}
# Delete anything which we might change between runs
vm_clean_caches() {
vm_cmd rm /ostree/repo/refs/heads/rpmostree/pkg/* -rf
}
# run rpm-ostree in vm
# - $@ args
vm_rpmostree() {
vm_cmd env ASAN_OPTIONS=detect_leaks=false rpm-ostree "$@"
}
# copy the test repo to the vm
# $1 - repo file mode: nogpgcheck (default), gpgcheck, skip (don't send)
vm_send_test_repo() {
mode=${1:-nogpgcheck}
# note we use -c here because we might be called twice within a second
vm_raw_rsync -c --delete ${test_tmpdir}/yumrepo $VM:/var/tmp/vmcheck
if [[ $mode == skip ]]; then
return
fi
cat > vmcheck.repo << EOF
[test-repo]
name=test-repo
baseurl=file:///var/tmp/vmcheck/yumrepo
EOF
if [[ $mode == gpgcheck ]]; then
cat >> vmcheck.repo <<EOF
gpgcheck=1
gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-fedora-25-primary
EOF
else
assert_streq "$mode" nogpgcheck
echo "Enabling vmcheck.repo without GPG"
echo 'gpgcheck=0' >> vmcheck.repo
fi
vm_send vmcheck.repo /etc/yum.repos.d
}
# wait until ssh is available on the vm
# - $1 timeout in second (optional)
# - $2 previous bootid (optional)
vm_ssh_wait() {
local timeout=${1:-0}; shift
local old_bootid=${1:-}; shift
if ! vm_cmd true; then
echo "Failed to log into VM, retrying with debug:"
$SSH -o LogLevel=debug true || true
fi
while [ $timeout -gt 0 ]; do
if bootid=$(vm_get_boot_id 2>/dev/null); then
if [[ $bootid != $old_bootid ]]; then
# if this is a reboot, display some info about new boot
if [ -n "$old_bootid" ]; then
vm_rpmostree status
vm_rpmostree --version
fi
return 0
fi
fi
if test $(($timeout % 5)) == 0; then
echo "Still failed to log into VM, retrying for $timeout seconds"
fi
timeout=$((timeout - 1))
sleep 1
done
false "Timed out while waiting for SSH."
}
vm_get_boot_id() {
vm_cmd cat /proc/sys/kernel/random/boot_id
}
# Run a command in the VM that will cause a reboot
vm_reboot_cmd() {
vm_cmd sync
local bootid=$(vm_get_boot_id 2>/dev/null)
vm_cmd "$@" || :
vm_ssh_wait 120 $bootid
}
# reboot the vm
vm_reboot() {
vm_reboot_cmd systemctl reboot
}
# check that the given files/dirs exist on the VM
# - $@ files/dirs to check for
vm_has_files() {
for file in "$@"; do
if ! vm_cmd test -e $file; then
return 1
fi
done
}
# check that the packages are installed
# - $@ packages to check for
vm_has_packages() {
for pkg in "$@"; do
if ! vm_cmd rpm -q $pkg; then
return 1
fi
done
}
# retrieve info from a deployment
# - $1 index of deployment (or -1 for booted)
# - $2 key to retrieve
vm_get_deployment_info() {
local idx=$1
local key=$2
vm_rpmostree status --json | \
python3 -c "
import sys, json
deployments = json.load(sys.stdin)[\"deployments\"]
idx = $idx
if idx < 0:
for i, depl in enumerate(deployments):
if depl[\"booted\"]:
idx = i
if idx < 0:
print(\"Failed to determine currently booted deployment\")
exit(1)
if idx >= len(deployments):
print(\"Deployment index $idx is out of range\")
exit(1)
depl = deployments[idx]
if \"$key\" in depl:
data = depl[\"$key\"]
if type(data) is list:
print(\" \".join(data))
else:
print(data)
"
}
# retrieve the deployment root
# - $1 index of deployment
vm_get_deployment_root() {
local idx=$1
local csum=$(vm_get_deployment_info $idx checksum)
local serial=$(vm_get_deployment_info $idx serial)
local osname=$(vm_get_deployment_info $idx osname)
echo /ostree/deploy/$osname/deploy/$csum.$serial
}
# retrieve info from the booted deployment
# - $1 key to retrieve
vm_get_booted_deployment_info() {
vm_get_deployment_info -1 $1
}
# print the layered packages
vm_get_layered_packages() {
vm_get_booted_deployment_info packages
}
# print the requested packages
vm_get_requested_packages() {
vm_get_booted_deployment_info requested-packages
}
vm_get_local_packages() {
vm_get_booted_deployment_info requested-local-packages
}
# check that the packages are currently layered
# - $@ packages to check for
vm_has_layered_packages() {
local pkgs=$(vm_get_layered_packages)
for pkg in "$@"; do
if [[ " $pkgs " != *$pkg* ]]; then
return 1
fi
done
}
# check that the packages are currently requested
# - $@ packages to check for
vm_has_requested_packages() {
local pkgs=$(vm_get_requested_packages)
for pkg in "$@"; do
if [[ " $pkgs " != *$pkg* ]]; then
return 1
fi
done
}
vm_has_local_packages() {
local pkgs=$(vm_get_local_packages)
for pkg in "$@"; do
if [[ " $pkgs " != *$pkg* ]]; then
return 1
fi
done
}
vm_has_dormant_packages() {
vm_has_requested_packages "$@" && \
! vm_has_layered_packages "$@"
}
vm_get_booted_stateroot() {
vm_get_booted_deployment_info osname
}
# retrieve the checksum of the currently booted deployment
vm_get_booted_csum() {
vm_get_booted_deployment_info checksum
}
# retrieve the checksum of the pending deployment
vm_get_pending_csum() {
vm_get_deployment_info 0 checksum
}
# make multiple consistency checks on a test pkg
# - $1 package to check for
# - $2 either "present" or "absent"
vm_assert_layered_pkg() {
local pkg=$1; shift
local policy=$1; shift
set +e
vm_has_packages $pkg; pkg_in_rpmdb=$?
vm_has_layered_packages $pkg; pkg_is_layered=$?
vm_has_local_packages $pkg; pkg_is_layered_local=$?
vm_has_requested_packages $pkg; pkg_is_requested=$?
[ $pkg_in_rpmdb == 0 ] && \
( ( [ $pkg_is_layered == 0 ] &&
[ $pkg_is_requested == 0 ] ) ||
[ $pkg_is_layered_local == 0 ] ); pkg_present=$?
[ $pkg_in_rpmdb != 0 ] && \
[ $pkg_is_layered != 0 ] && \
[ $pkg_is_layered_local != 0 ] && \
[ $pkg_is_requested != 0 ]; pkg_absent=$?
set -e
if [ $policy == present ] && [ $pkg_present != 0 ]; then
vm_cmd rpm-ostree status
assert_not_reached "pkg $pkg is not present"
fi
if [ $policy == absent ] && [ $pkg_absent != 0 ]; then
vm_cmd rpm-ostree status
assert_not_reached "pkg $pkg is not absent"
fi
}
# Takes a list of `jq` expressions, each of which should evaluate to a boolean,
# and asserts that they are true.
vm_assert_status_jq() {
vm_rpmostree status --json > status.json
assert_jq status.json "$@"
}
vm_pending_is_staged() {
vm_rpmostree status --json > status-staged.json
local rc=1
if jq -e ".deployments[0][\"staged\"]" < status-staged.json; then
rc=0
fi
rm -f status-staged.json
return $rc
}
# Like build_rpm, but also sends it to the VM
vm_build_rpm() {
build_rpm "$@"
vm_send_test_repo
}
# Like uinfo_cmd, but also sends it to the VM
vm_uinfo() {
uinfo_cmd "$@"
vm_send_test_repo
}
# Like vm_build_rpm but takes a yumrepo mode
vm_build_rpm_repo_mode() {
mode=$1; shift
build_rpm "$@"
vm_send_test_repo $mode
}
vm_build_selinux_rpm() {
build_selinux_rpm "$@"
vm_send_test_repo
}
vm_get_journal_cursor() {
vm_cmd journalctl -o json -n 1 | jq -r '.["__CURSOR"]'
}
# Wait for a message logged after $cursor matching a regexp to appear
# $1 - cursor
# $2 - regex to wait for
vm_wait_content_after_cursor() {
from_cursor=$1; shift
regex=$1; shift
vm_shell_inline <<EOF
tmpf=\$(mktemp /var/tmp/journal.XXXXXX)
for x in \$(seq 60); do
journalctl -u rpm-ostreed --after-cursor "${from_cursor}" > \${tmpf}
if grep -q -e "${regex}" \${tmpf}; then
exit 0
else
cat \${tmpf}
sleep 1
fi
done
echo "timed out after 60s" 1>&2
journalctl -u rpm-ostreed --after-cursor "${from_cursor}" | tail -100
exit 1
EOF
}
# Minor helper that makes sure to get quoting right
vm_get_journal_after_cursor() {
from_cursor=$1; shift
to_file=$1; shift
# add an extra helping of quotes for hungry ssh
vm_cmd journalctl --after-cursor "'$from_cursor'" > $to_file
}
vm_assert_journal_has_content() {
from_cursor=$1; shift
vm_get_journal_after_cursor $from_cursor tmp-journal.txt
assert_file_has_content tmp-journal.txt "$@"
rm -f tmp-journal.txt
}
# usage: <podman args> -- <container args>
vm_run_container() {
local podman_args=
while [ $# -ne 0 ]; do
local arg=$1; shift
if [[ $arg == -- ]]; then
break
fi
podman_args="$podman_args $arg"
done
[ $# -ne 0 ] || fatal "No container args provided"
# just automatically always share dnf cache so we don't redownload each time
# (use -n so this ssh invocation doesn't consume stdin)
vm_cmd -n mkdir -p /var/cache/dnf
vm_cmd podman run --rm -v /var/cache/dnf:/var/cache/dnf:z $podman_args \
registry.fedoraproject.org/fedora:30 "$@"
}
# $1 - service name
# $2 - dir to serve
# $3 - port to serve on
vm_start_httpd() {
local name=$1; shift
local dir=$1; shift
local port=$1; shift
vm_cmd podman rm -f $name || true
vm_run_container --net=host -d --name $name --privileged \
-v $dir:/srv --workdir /srv -- \
python3 -m http.server $port
# NB: the EXIT trap is used by libtest, but not the ERR trap
trap "vm_stop_httpd $name" ERR
set -E # inherit trap
# Ideally systemd-run would support .socket units or something
vm_cmd 'while ! curl --head http://127.0.0.1:8888 &>/dev/null; do sleep 1; done'
}
# $1 - service name
vm_stop_httpd() {
local name=$1; shift
vm_cmd podman rm -f $name
set +E
trap - ERR
}
# start up an ostree server to be used as an http remote
vm_ostreeupdate_prepare_repo() {
# Really testing this like a user requires a remote ostree server setup.
# Let's start by setting up the repo.
REMOTE_OSTREE=/ostree/repo/tmp/vmcheck-remote
vm_cmd mkdir -p $REMOTE_OSTREE
vm_cmd ostree init --repo=$REMOTE_OSTREE --mode=archive
vm_start_httpd ostree_server $REMOTE_OSTREE 8888
}
# this is split out for the sole purpose of making iterating easier when hacking
# (see below for more details)
_init_updated_rpmmd_repo() {
vm_build_rpm base-pkg-foo version 1.4 release 8 # upgraded
vm_build_rpm base-pkg-bar version 0.9 release 3 # downgraded
vm_build_rpm base-pkg-boo version 3.7 release 2.11 # added
vm_uinfo add VMCHECK-ENH enhancement
vm_uinfo add VMCHECK-SEC-NONE security none
vm_uinfo add VMCHECK-SEC-LOW security low
vm_uinfo add VMCHECK-SEC-CRIT security critical
vm_build_rpm base-pkg-enh version 2.0 uinfo VMCHECK-ENH
vm_build_rpm base-pkg-sec-none version 2.0 uinfo VMCHECK-SEC-NONE
vm_build_rpm base-pkg-sec-low version 2.0 uinfo VMCHECK-SEC-LOW
vm_build_rpm base-pkg-sec-crit version 2.0 uinfo VMCHECK-SEC-CRIT
vm_uinfo add-ref VMCHECK-SEC-LOW 1 http://example.com/vuln1 "CVE-12-34 vuln1"
vm_uinfo add-ref VMCHECK-SEC-LOW 2 http://example.com/vuln2 "CVE-12-34 vuln2"
vm_uinfo add-ref VMCHECK-SEC-LOW 3 http://example.com/vuln3 "CVE-56-78 CVE-90-12 vuln3"
vm_uinfo add-ref VMCHECK-SEC-LOW 4 http://example.com/vuln4 "CVE-12-JUNK CVE-JUNK vuln4"
}
# Start up a remote, and create two new commits (v1 and v2) which contain new
# pkgs. The 'vmcheck' ref on the remote is set at v1. You can then make a new
# update appear later using "vm_ostreeupdate_create v2".
vm_ostreeupdate_prepare() {
# first, let's make sure the timer is disabled so it doesn't mess up with our
# tests
vm_cmd systemctl disable --now rpm-ostreed-automatic.timer
# Prepare an OSTree repo with updates
vm_ostreeupdate_prepare_repo
# (delete ref but don't prune for easier debugging)
vm_cmd ostree refs --repo=$REMOTE_OSTREE vmcheck --delete
# now let's build some pkgs that we'll jury-rig into a base update
# this whole block can be commented out (except the init_updated_rpmmd_repo
# call) after the first run for a speed-up when iterating locally
vm_build_rpm base-pkg-foo version 1.4 release 7
vm_build_rpm base-pkg-bar
vm_build_rpm base-pkg-baz version 1.1 release 1
vm_build_rpm base-pkg-enh
vm_build_rpm base-pkg-sec-none
vm_build_rpm base-pkg-sec-low
vm_build_rpm base-pkg-sec-crit
vm_rpmostree install base-pkg-{foo,bar,baz,enh,sec-{none,low,crit}}
vm_ostreeupdate_lift_commit $(vm_get_pending_csum) v1
vm_rpmostree cleanup -p
# ok, we don't need those RPMs anymore since they're part of the base
rm -rf $test_tmpdir/yumrepo
# create new versions of those RPMs that we install in v2; we keep the repo
# around since that's where e.g. advisories are stored too when analyzing
# the v2 ostree update
_init_updated_rpmmd_repo
vm_rpmostree install base-pkg-{foo,bar,boo,enh,sec-{none,low,crit}}
vm_ostreeupdate_lift_commit $(vm_get_pending_csum) v2
vm_rpmostree cleanup -p
vm_ostreeupdate_create v1
vm_cmd ostree remote add vmcheckmote --no-gpg-verify http://localhost:8888/
vm_rpmostree reload
}
vm_ostreeupdate_prepare_reboot() {
vm_ostreeupdate_prepare
vm_rpmostree rebase vmcheckmote:vmcheck
vm_reboot
vm_rpmostree cleanup -pr
vm_assert_status_jq ".deployments[0][\"origin\"] == \"vmcheckmote:vmcheck\"" \
".deployments[0][\"booted\"]" \
".deployments[0][\"version\"] == \"v1\""
vm_rpmostree status > status.txt
assert_file_has_content_literal status.txt 'AutomaticUpdates: disabled'
# start it up again since we rebooted
vm_start_httpd ostree_server $REMOTE_OSTREE 8888
}
vm_change_update_policy() {
policy=$1; shift
vm_shell_inline <<EOF
cp /usr/etc/rpm-ostreed.conf /etc
echo -e "[Daemon]\nAutomaticUpdatePolicy=$policy" > /etc/rpm-ostreed.conf
rpm-ostree reload
EOF
}
# APIs to build up a history on the server. Rather than wasting time
# composing trees for real, we just use client package layering to create new
# trees that we then "lift" into the server before cleaning them up client-side.
# steal a commit from the system repo and tag it as a new version
vm_ostreeupdate_lift_commit() {
checksum=$1; shift
# ostree doesn't support tags, so just shove it in a branch
branch=vmcheck_tmp/$1; shift
vm_cmd ostree pull-local --repo=$REMOTE_OSTREE --disable-fsync \
/ostree/repo $checksum
vm_cmd ostree --repo=$REMOTE_OSTREE refs $branch --delete
vm_cmd ostree --repo=$REMOTE_OSTREE refs $checksum --create=$branch
}
_commit_and_inject_pkglist() {
local version=$1; shift
local src_ref=$1; shift
vm_cmd ostree commit --repo=$REMOTE_OSTREE -b vmcheck --fsync=no \
--tree=ref=$src_ref --add-metadata-string=version=$version
vm_rpmostree testutils inject-pkglist $REMOTE_OSTREE vmcheck
}
# use a previously stolen commit to create an update on our vmcheck branch,
# complete with version string and pkglist metadata
vm_ostreeupdate_create() {
version=$1; shift
_commit_and_inject_pkglist $version vmcheck_tmp/$version
}
# create a new no-op update with version metadata $1
vm_ostreeupdate_create_noop() {
version=$1; shift
_commit_and_inject_pkglist $version vmcheck
}
# takes a layered commit, and makes it into a base
vm_ostree_repo_commit_layered_as_base() {
local repo=$1; shift
local from_rev=$1; shift
local to_ref=$1; shift
local d=$repo/tmp/vmcheck_commit.tmp
rm -rf $d
vm_cmd ostree checkout --repo=$repo -H --fsync=no $from_rev $d
# need to update the base rpmdb
vm_cmd mkdir -p $d/usr/lib/sysimage/rpm-ostree-base-db
vm_cmd rsync -qa --delete $d/usr/share/rpm/ $d/usr/lib/sysimage/rpm-ostree-base-db
vm_cmd ostree commit --repo=$repo -b $to_ref --link-checkout-speedup --fsync=no --consume $d
# and inject pkglist metadata
vm_rpmostree testutils inject-pkglist $repo $to_ref >/dev/null
}
vm_ostree_commit_layered_as_base() {
vm_ostree_repo_commit_layered_as_base /ostree/repo "$@"
}
vm_status_watch_start() {
rm -rf status-watch.txt
while sleep 1; do
vm_rpmostree status >> status-watch.txt
done &
_status_watch_pid=$!
# NB: the EXIT trap is used by libtest, but not the ERR trap
trap "kill $_status_watch_pid" ERR
set -E # inherit trap
}
vm_status_watch_check() {
[ -n "${_status_watch_pid:-}" ]
kill $_status_watch_pid
_status_watch_pid=
set +E
[ -f status-watch.txt ]
assert_file_has_content_literal status-watch.txt "$@"
rm -rf status-watch.txt
}