podman/test/system/120-load.bats
Ed Santiago e4dcb1004f System tests: fix a multiarch problem
The multi-image load test was failing on non-x86_64 arch,
because the images used by the test (:00000000, :20200902)
did not have manifests for the given arch.

Solution: all we need are two nonlocal images. Use the
predefined NONLOCAL_IMAGE for one, and a new :multiimage
tag (manually created, currently == :20210610) for the other.
Document, so if/when RHEL adds new supported arches, the
test will fail but a maintainer will have a clue what to do.

Also, as long as I'm in here: add 'image prune -f' to teardown
in build.bats, to avoid seeing lots of red "stray image"
warnings in test logs.

Also: skip a broken/flaky python test

Signed-off-by: Ed Santiago <santiago@redhat.com>
2021-07-15 11:29:58 -06:00

187 lines
6.3 KiB
Bash

#!/usr/bin/env bats -*- bats -*-
#
# tests for podman load
#
load helpers
# Custom helpers for this test only. These just save us having to duplicate
# the same thing four times (two tests, each with -i and stdin).
#
# initialize, read image ID and name
get_iid_and_name() {
run_podman images -a --format '{{.ID}} {{.Repository}}:{{.Tag}}'
read iid img_name < <(echo "$output")
archive=$PODMAN_TMPDIR/myimage-$(random_string 8).tar
}
# Simple verification of image ID and name
verify_iid_and_name() {
run_podman images -a --format '{{.ID}} {{.Repository}}:{{.Tag}}'
read new_iid new_img_name < <(echo "$output")
# Verify
is "$new_iid" "$iid" "Image ID of loaded image == original"
is "$new_img_name" "$1" "Name & tag of restored image"
}
@test "podman load invalid file" {
# Regression test for #9672 to make sure invalid input yields errors.
invalid=$PODMAN_TMPDIR/invalid
echo "I am an invalid file and should cause a podman-load error" > $invalid
run_podman 125 load -i $invalid
# podman and podman-remote emit different messages; this is a common string
is "$output" ".*payload does not match any of the supported image formats .*" \
"load -i INVALID fails with expected diagnostic"
}
@test "podman save to pipe and load" {
# Generate a random name and tag (must be lower-case)
local random_name=x0$(random_string 12 | tr A-Z a-z)
local random_tag=t0$(random_string 7 | tr A-Z a-z)
local fqin=localhost/$random_name:$random_tag
run_podman tag $IMAGE $fqin
# Believe it or not, 'podman load' would barf if any path element
# included a capital letter
archive=$PODMAN_TMPDIR/MySubDirWithCaps/MyImage-$(random_string 8).tar
mkdir -p $(dirname $archive)
# We can't use run_podman because that uses the BATS 'run' function
# which redirects stdout and stderr. Here we need to guarantee
# that podman's stdout is a pipe, not any other form of redirection
$PODMAN save --format oci-archive $fqin | cat >$archive
if [ "$status" -ne 0 ]; then
die "Command failed: podman save ... | cat"
fi
# Make sure we can reload it
run_podman rmi $fqin
run_podman load -i $archive
# FIXME: cannot compare IID, see #7371, so we check only the tag
run_podman images $fqin --format '{{.Repository}}:{{.Tag}}'
is "$output" "$fqin" "image preserves name across save/load"
# Load with a new tag
local new_name=x1$(random_string 14 | tr A-Z a-z)
local new_tag=t1$(random_string 6 | tr A-Z a-z)
run_podman rmi $fqin
run_podman load -i $archive
run_podman images --format '{{.Repository}}:{{.Tag}}' --sort tag
is "${lines[0]}" "$IMAGE" "image is preserved"
is "${lines[1]}" "$fqin" "image is reloaded with old fqin"
# Clean up
run_podman rmi $fqin
}
@test "podman load - by image ID" {
# FIXME: how to build a simple archive instead?
get_iid_and_name
# Save image by ID, and remove it.
run_podman save $iid -o $archive
run_podman rmi $iid
# Load using -i; IID should be preserved, but name is not.
run_podman load -i $archive
verify_iid_and_name "<none>:<none>"
# Same as above, using stdin
run_podman rmi $iid
run_podman load < $archive
verify_iid_and_name "<none>:<none>"
# Same as above, using stdin but with `podman image load`
run_podman rmi $iid
run_podman image load < $archive
verify_iid_and_name "<none>:<none>"
# Cleanup: since load-by-iid doesn't preserve name, re-tag it;
# otherwise our global teardown will rmi and re-pull our standard image.
run_podman tag $iid $img_name
}
@test "podman load - by image name" {
get_iid_and_name
run_podman save $img_name -o $archive
run_podman rmi $iid
# Load using -i; this time the image should be tagged.
run_podman load -i $archive
verify_iid_and_name $img_name
run_podman rmi $iid
# Also make sure that `image load` behaves the same.
run_podman image load -i $archive
verify_iid_and_name $img_name
run_podman rmi $iid
# Same as above, using stdin
run_podman load < $archive
verify_iid_and_name $img_name
}
@test "podman load - redirect corrupt payload" {
run_podman 125 load <<< "Danger, Will Robinson!! This is a corrupt tarball!"
is "$output" \
".*payload does not match any of the supported image formats .*" \
"Diagnostic from 'podman load' unknown/corrupt payload"
}
@test "podman load - multi-image archive" {
# img1 & 2 should be images that are not locally present; they must also
# be usable on the host arch. The nonlocal image (:000000xx) is kept
# up-to-date for all RHEL/Fedora arches; the other image we use is
# the one tagged ':multiimage', which as of 2021-07-15 is :20210610
# but that tag will grow stale over time. If/when this test fails,
# your first approach should be to manually update :multiimage to
# point to a more recent testimage. (Use the quay.io GUI, it's waaay
# easier than pulling/pushing the correct manifest.)
img1=${PODMAN_NONLOCAL_IMAGE_FQN}
img2="$PODMAN_TEST_IMAGE_REGISTRY/$PODMAN_TEST_IMAGE_USER/$PODMAN_TEST_IMAGE_NAME:multiimage"
archive=$PODMAN_TMPDIR/myimage-$(random_string 8).tar
run_podman pull $img1
run_podman pull $img2
run_podman save -m -o $archive $img1 $img2
run_podman rmi -f $img1 $img2
run_podman load -i $archive
run_podman image exists $img1
run_podman image exists $img2
run_podman rmi -f $img1 $img2
}
@test "podman load - multi-image archive with redirect" {
# (see comments in test above re: img1 & 2)
img1=${PODMAN_NONLOCAL_IMAGE_FQN}
img2="$PODMAN_TEST_IMAGE_REGISTRY/$PODMAN_TEST_IMAGE_USER/$PODMAN_TEST_IMAGE_NAME:multiimage"
archive=$PODMAN_TMPDIR/myimage-$(random_string 8).tar
run_podman pull $img1
run_podman pull $img2
# We can't use run_podman because that uses the BATS 'run' function
# which redirects stdout and stderr. Here we need to guarantee
# that podman's stdout is a pipe, not any other form of redirection
$PODMAN save -m $img1 $img2 | cat >$archive
if [ "$status" -ne 0 ]; then
die "Command failed: podman save ... | cat"
fi
run_podman rmi -f $img1 $img2
run_podman load -i $archive
run_podman image exists $img1
run_podman image exists $img2
run_podman rmi -f $img1 $img2
}
# vim: filetype=sh