system/systemd
system/systemd
1
0
Fork 0
mirror of https://github.com/systemd/systemd synced 2024-09-06 16:56:43 +00:00
Code Issues Packages Projects Releases Wiki Activity
systemd/test/units/testsuite-15.sh

581 lines
22 KiB
Bash
Raw Normal View History

tests: use !/usr/bin/env bash consistently This adjustment was last done in f96bc66901ecbcf7ecb280d9442fd0cc92e67186, but it seems some files were missed and some were added later.
2021-10-17 16:14:58 +00:00
#!/usr/bin/env bash
tests: add spdx headers to scripts and Makefiles
2021-10-17 16:13:06 +00:00
# SPDX-License-Identifier: LGPL-2.1-or-later
test: use set -eux and set -o pipefail everywhere This should make the scripts more robust.
2021-04-09 17:39:41 +00:00
set -eux
set -o pipefail
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_unit () {
local UNIT_NAME="${1:?}"
systemctl stop "$UNIT_NAME" 2>/dev/null || :
rm -f /{etc,run,usr/lib}/systemd/system/"$UNIT_NAME"
rm -fr /{etc,run,usr/lib}/systemd/system/"$UNIT_NAME".d
rm -fr /{etc,run,usr/lib}/systemd/system/"$UNIT_NAME".{wants,requires}
if [[ $UNIT_NAME == *@ ]]; then
local base="${UNIT_NAME%@*}"
local suffix="${UNIT_NAME##*.}"
systemctl stop "$base@"*."$suffix" 2>/dev/null || :
rm -f /{etc,run,usr/lib}/systemd/system/"$base@"*."$suffix"
rm -fr /{etc,run,usr/lib}/systemd/system/"$base@"*."$suffix".d
rm -fr /{etc,run,usr/lib}/systemd/system/"$base@"*."$suffix".{wants,requires}
test/TEST-15: remove all created unit files We would miss anything created under a template instance.
2020-03-21 15:30:27 +00:00
fi
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
}
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units () {
test: use quotes where necessary to avoid possible word splitting.
2021-04-09 17:49:32 +00:00
for u in "$@"; do
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_unit "$u"
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
done
systemctl daemon-reload
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
}
create_service () {
tree-wide: drop manually-crafted message for missing variables Bash will generate a very nice message for us: /tmp/ff.sh: line 1: SOMEVAR: parameter null or not set Let's save some keystrokes by not replacing this with our own inferior messages.
2022-05-04 06:24:06 +00:00
local SERVICE_NAME="${1:?}"
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units "${SERVICE_NAME}".service
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
test: use quotes where necessary to avoid possible word splitting.
2021-04-09 17:49:32 +00:00
cat >/etc/systemd/system/"$SERVICE_NAME".service <<EOF
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
[Unit]
test: use quotes where necessary to avoid possible word splitting.
2021-04-09 17:49:32 +00:00
Description=$SERVICE_NAME unit
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
[Service]
TEST-15: use ${:?} and shorten things a bit
2021-04-29 07:20:39 +00:00
ExecStart=sleep 100000
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
EOF
TEST-15: use ${:?} and shorten things a bit
2021-04-29 07:20:39 +00:00
mkdir -p /{etc,run,usr/lib}/systemd/system/"$SERVICE_NAME".service.{d,wants,requires}
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
}
create_services () {
test: use quotes where necessary to avoid possible word splitting.
2021-04-09 17:49:32 +00:00
for u in "$@"; do
create_service "$u"
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
done
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
}
check_ok () {
TEST-15: use ${:?} and shorten things a bit
2021-04-29 07:20:39 +00:00
x="$(systemctl show --value -p "${2:?}" "${1:?}")"
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
case "$x" in
TEST-15: use ${:?} and shorten things a bit
2021-04-29 07:20:39 +00:00
*${3:?}*) return 0 ;;
*) return 1 ;;
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
esac
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
}
check_ko () {
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
! check_ok "$@"
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
}
test_basic_dropins () {
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
echo "Testing basic dropins..."
echo "*** test a wants b wants c"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b test15-c
ln -s ../test15-b.service /etc/systemd/system/test15-a.service.wants/
ln -s ../test15-c.service /etc/systemd/system/test15-b.service.wants/
check_ok test15-a Wants test15-b.service
check_ok test15-b Wants test15-c.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
echo "*** test a wants,requires b"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b test15-c
ln -s ../test15-b.service /etc/systemd/system/test15-a.service.wants/
ln -s ../test15-b.service /etc/systemd/system/test15-a.service.requires/
check_ok test15-a Wants test15-b.service
check_ok test15-a Requires test15-b.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
echo "*** test a wants nonexistent"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_service test15-a
ln -s ../nonexistent.service /etc/systemd/system/test15-a.service.wants/
check_ok test15-a Wants nonexistent.service
systemctl start test15-a
systemctl stop test15-a
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
echo "*** test a requires nonexistent"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
ln -sf ../nonexistent.service /etc/systemd/system/test15-a.service.requires/
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
systemctl daemon-reload
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
check_ok test15-a Requires nonexistent.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'b' is already loaded when 'c' pulls it in via a dropin.
echo "*** test a,c require b"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b test15-c
ln -sf ../test15-b.service /etc/systemd/system/test15-a.service.requires/
ln -sf ../test15-b.service /etc/systemd/system/test15-c.service.requires/
systemctl start test15-a
check_ok test15-c Requires test15-b.service
systemctl stop test15-a test15-b
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'b' is already loaded when 'c' pulls it in via an alias dropin.
echo "*** test a wants alias"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b test15-c
ln -sf test15-c.service /etc/systemd/system/test15-c1.service
ln -sf ../test15-c.service /etc/systemd/system/test15-a.service.wants/
ln -sf ../test15-c1.service /etc/systemd/system/test15-b.service.wants/
systemctl start test15-a
check_ok test15-a Wants test15-c.service
check_ok test15-b Wants test15-c.service
systemctl stop test15-a test15-c
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
core: change top-level drop-in from -.service.d to service.d Discussed in #13743, the -.service semantic conflicts with the existing root mount and slice names, making this feature not uniformly extensible to all types. Change the name to be <type>.d instead. Updating to this format also extends the top-level dropin to unit types.
2019-11-05 02:29:55 +00:00
echo "*** test service.d/ top level drop-in"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b
check_ko test15-a ExecCondition "/bin/echo a"
check_ko test15-b ExecCondition "/bin/echo b"
core: change top-level drop-in from -.service.d to service.d Discussed in #13743, the -.service semantic conflicts with the existing root mount and slice names, making this feature not uniformly extensible to all types. Change the name to be <type>.d instead. Updating to this format also extends the top-level dropin to unit types.
2019-11-05 02:29:55 +00:00
mkdir -p /usr/lib/systemd/system/service.d
cat >/usr/lib/systemd/system/service.d/override.conf <<EOF
shared/dropin: support -.service.d/ top level drop-in for service units Closes #12830
2019-10-05 00:39:34 +00:00
[Service]
ExecCondition=/bin/echo %n
EOF
test/TEST-15: call daemon-reload in two more places Sometimes the test would fail there, nondeterministically. I'm not sure why, but relying on PID1 not caching the file is clearly very brittle. Let's instead call daemon-reload.
2020-03-21 16:01:09 +00:00
systemctl daemon-reload
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
check_ok test15-a ExecCondition "/bin/echo test15-a"
check_ok test15-b ExecCondition "/bin/echo test15-b"
core: change top-level drop-in from -.service.d to service.d Discussed in #13743, the -.service semantic conflicts with the existing root mount and slice names, making this feature not uniformly extensible to all types. Change the name to be <type>.d instead. Updating to this format also extends the top-level dropin to unit types.
2019-11-05 02:29:55 +00:00
rm -rf /usr/lib/systemd/system/service.d
shared/dropin: support -.service.d/ top level drop-in for service units Closes #12830
2019-10-05 00:39:34 +00:00
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units test15-{a,b,c,c1}.service
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
}
TEST-15-DROPINS: add test for linked units
2021-02-14 14:08:49 +00:00
test_linked_units () {
echo "Testing linked units..."
echo "*** test linked unit (same basename)"
create_service test15-a
mv /etc/systemd/system/test15-a.service /
ln -s /test15-a.service /etc/systemd/system/
ln -s test15-a.service /etc/systemd/system/test15-b.service
check_ok test15-a Names test15-a.service
check_ok test15-a Names test15-b.service
echo "*** test linked unit (cross basename)"
mv /test15-a.service /test15-a@.scope
ln -fs /test15-a@.scope /etc/systemd/system/test15-a.service
systemctl daemon-reload
check_ok test15-a Names test15-a.service
check_ok test15-a Names test15-b.service
TEST-15-DROPINS: improve check https://github.com/systemd/systemd/pull/18579#discussion_r588983813
2021-03-07 11:27:55 +00:00
check_ko test15-a Names test15-a@ # test15-a@.scope is the symlink target.
# Make sure it is completely ignored.
TEST-15-DROPINS: add test for linked units
2021-02-14 14:08:49 +00:00
rm /test15-a@.scope
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units test15-{a,b}.service
TEST-15-DROPINS: add test for linked units
2021-02-14 14:08:49 +00:00
}
TEST-15: add test for template aliases
2021-04-29 07:47:23 +00:00
test_template_alias() {
echo "Testing instance alias..."
echo "*** forward"
create_service test15-a@
ln -s test15-a@inst.service /etc/systemd/system/test15-b@inst.service # alias
check_ok test15-a@inst Names test15-a@inst.service
check_ok test15-a@inst Names test15-b@inst.service
check_ok test15-a@other Names test15-a@other.service
check_ko test15-a@other Names test15-b@other.service
echo "*** reverse"
systemctl daemon-reload
check_ok test15-b@inst Names test15-a@inst.service
check_ok test15-b@inst Names test15-b@inst.service
check_ko test15-b@other Names test15-a@other.service
check_ok test15-b@other Names test15-b@other.service
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units test15-{a,b}@.service
TEST-15: add test for template aliases
2021-04-29 07:47:23 +00:00
}
TEST-15: also test hierarchical drop-ins for slices Slices are worth testing too, because they don't need a fragment path so they behave slightly differently than service units. I'm making this a separate patch from the actual tests that I wanted to add later because it's complex enough on its own.
2022-10-16 10:54:34 +00:00
test_hierarchical_service_dropins () {
echo "Testing hierarchical service dropins..."
Add failing test to show service.d global drop-in does not get overridden by more specific dropins
2019-12-27 14:36:49 +00:00
echo "*** test service.d/ top level drop-in"
create_services a-b-c
check_ko a-b-c ExecCondition "/bin/echo service.d"
check_ko a-b-c ExecCondition "/bin/echo a-.service.d"
check_ko a-b-c ExecCondition "/bin/echo a-b-.service.d"
check_ko a-b-c ExecCondition "/bin/echo a-b-c.service.d"
for dropin in service.d a-.service.d a-b-.service.d a-b-c.service.d; do
mkdir -p /usr/lib/systemd/system/$dropin
echo "
[Service]
ExecCondition=/bin/echo $dropin
TEST-*: use spacing before redirection operator, but not after << EOF → <<EOF > foo < bar → >foo <bar
2021-04-07 22:09:55 +00:00
" >/usr/lib/systemd/system/$dropin/override.conf
test/TEST-15: call daemon-reload in two more places Sometimes the test would fail there, nondeterministically. I'm not sure why, but relying on PID1 not caching the file is clearly very brittle. Let's instead call daemon-reload.
2020-03-21 16:01:09 +00:00
systemctl daemon-reload
Add failing test to show service.d global drop-in does not get overridden by more specific dropins
2019-12-27 14:36:49 +00:00
check_ok a-b-c ExecCondition "/bin/echo $dropin"
done
for dropin in service.d a-.service.d a-b-.service.d a-b-c.service.d; do
rm -rf /usr/lib/systemd/system/$dropin
done
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units a-b-c.service
Add failing test to show service.d global drop-in does not get overridden by more specific dropins
2019-12-27 14:36:49 +00:00
}
TEST-15: also test hierarchical drop-ins for slices Slices are worth testing too, because they don't need a fragment path so they behave slightly differently than service units. I'm making this a separate patch from the actual tests that I wanted to add later because it's complex enough on its own.
2022-10-16 10:54:34 +00:00
test_hierarchical_slice_dropins () {
echo "Testing hierarchical slice dropins..."
echo "*** test slice.d/ top level drop-in"
# Slice units don't even need a fragment, so we test the defaults here
check_ok a-b-c.slice Description "Slice /a/b/c"
check_ok a-b-c.slice MemoryMax "infinity"
# Test drop-ins
for dropin in slice.d a-.slice.d a-b-.slice.d a-b-c.slice.d; do
mkdir -p /usr/lib/systemd/system/$dropin
echo "
[Slice]
MemoryMax=1000000000
" >/usr/lib/systemd/system/$dropin/override.conf
systemctl daemon-reload
check_ok a-b-c.slice MemoryMax "1000000000"
rm /usr/lib/systemd/system/$dropin/override.conf
done
# Test unit with a fragment
echo "
[Slice]
MemoryMax=1000000001
" >/usr/lib/systemd/system/a-b-c.slice
systemctl daemon-reload
check_ok a-b-c.slice MemoryMax "1000000001"
clear_units a-b-c.slice
}
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
test_template_dropins () {
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
echo "Testing template dropins..."
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
create_services foo bar@ yup@
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
# Declare some deps to check if the body was loaded
cat >>/etc/systemd/system/bar@.service <<EOF
[Unit]
After=bar-template-after.device
EOF
cat >>/etc/systemd/system/yup@.service <<EOF
[Unit]
After=yup-template-after.device
EOF
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
ln -s /etc/systemd/system/bar@.service /etc/systemd/system/foo.service.wants/bar@1.service
check_ok foo Wants bar@1.service
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
echo "*** test bar-alias@.service→bar@.service, but instance symlinks point to yup@.service ***"
ln -s bar@.service /etc/systemd/system/bar-alias@.service
ln -s bar@1.service /etc/systemd/system/bar-alias@1.service
ln -s yup@.service /etc/systemd/system/bar-alias@2.service
ln -s yup@3.service /etc/systemd/system/bar-alias@3.service
# create some dropin deps
mkdir -p /etc/systemd/system/bar@{,0,1,2,3}.service.requires/
mkdir -p /etc/systemd/system/yup@{,0,1,2,3}.service.requires/
mkdir -p /etc/systemd/system/bar-alias@{,0,1,2,3}.service.requires/
ln -s ../bar-template-requires.device /etc/systemd/system/bar@.service.requires/
ln -s ../bar-0-requires.device /etc/systemd/system/bar@0.service.requires/
ln -s ../bar-1-requires.device /etc/systemd/system/bar@1.service.requires/
ln -s ../bar-2-requires.device /etc/systemd/system/bar@2.service.requires/
ln -s ../bar-3-requires.device /etc/systemd/system/bar@3.service.requires/
ln -s ../yup-template-requires.device /etc/systemd/system/yup@.service.requires/
ln -s ../yup-0-requires.device /etc/systemd/system/yup@0.service.requires/
ln -s ../yup-1-requires.device /etc/systemd/system/yup@1.service.requires/
ln -s ../yup-2-requires.device /etc/systemd/system/yup@2.service.requires/
ln -s ../yup-3-requires.device /etc/systemd/system/yup@3.service.requires/
ln -s ../bar-alias-template-requires.device /etc/systemd/system/bar-alias@.service.requires/
ln -s ../bar-alias-0-requires.device /etc/systemd/system/bar-alias@0.service.requires/
ln -s ../bar-alias-1-requires.device /etc/systemd/system/bar-alias@1.service.requires/
ln -s ../bar-alias-2-requires.device /etc/systemd/system/bar-alias@2.service.requires/
ln -s ../bar-alias-3-requires.device /etc/systemd/system/bar-alias@3.service.requires/
systemctl daemon-reload
echo '*** bar@0 is aliased by bar-alias@0 ***'
systemctl show -p Names,Requires bar@0
systemctl show -p Names,Requires bar-alias@0
check_ok bar@0 Names bar@0
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
check_ok bar@0 Names bar-alias@0
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
check_ok bar@0 After bar-template-after.device
check_ok bar@0 Requires bar-0-requires.device
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
check_ok bar@0 Requires bar-alias-0-requires.device
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
check_ok bar@0 Requires bar-template-requires.device
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
check_ok bar@0 Requires bar-alias-template-requires.device
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
check_ko bar@0 Requires yup-template-requires.device
check_ok bar-alias@0 After bar-template-after.device
check_ok bar-alias@0 Requires bar-0-requires.device
check_ok bar-alias@0 Requires bar-alias-0-requires.device
check_ok bar-alias@0 Requires bar-template-requires.device
check_ok bar-alias@0 Requires bar-alias-template-requires.device
check_ko bar-alias@0 Requires yup-template-requires.device
check_ko bar-alias@0 Requires yup-0-requires.device
echo '*** bar@1 is aliased by bar-alias@1 ***'
systemctl show -p Names,Requires bar@1
systemctl show -p Names,Requires bar-alias@1
check_ok bar@1 Names bar@1
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
check_ok bar@1 Names bar-alias@1
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
check_ok bar@1 After bar-template-after.device
check_ok bar@1 Requires bar-1-requires.device
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
check_ok bar@1 Requires bar-alias-1-requires.device
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
check_ok bar@1 Requires bar-template-requires.device
# See https://github.com/systemd/systemd/pull/13119#discussion_r308145418
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
check_ok bar@1 Requires bar-alias-template-requires.device
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
check_ko bar@1 Requires yup-template-requires.device
check_ko bar@1 Requires yup-1-requires.device
check_ok bar-alias@1 After bar-template-after.device
check_ok bar-alias@1 Requires bar-1-requires.device
check_ok bar-alias@1 Requires bar-alias-1-requires.device
check_ok bar-alias@1 Requires bar-template-requires.device
check_ok bar-alias@1 Requires bar-alias-template-requires.device
check_ko bar-alias@1 Requires yup-template-requires.device
check_ko bar-alias@1 Requires yup-1-requires.device
echo '*** bar-alias@2 aliases yup@2, bar@2 is independent ***'
systemctl show -p Names,Requires bar@2
systemctl show -p Names,Requires bar-alias@2
check_ok bar@2 Names bar@2
check_ko bar@2 Names bar-alias@2
check_ok bar@2 After bar-template-after.device
check_ok bar@2 Requires bar-2-requires.device
check_ko bar@2 Requires bar-alias-2-requires.device
check_ok bar@2 Requires bar-template-requires.device
check_ko bar@2 Requires bar-alias-template-requires.device
check_ko bar@2 Requires yup-template-requires.device
check_ko bar@2 Requires yup-2-requires.device
check_ko bar-alias@2 After bar-template-after.device
check_ko bar-alias@2 Requires bar-2-requires.device
check_ok bar-alias@2 Requires bar-alias-2-requires.device
check_ko bar-alias@2 Requires bar-template-requires.device
check_ok bar-alias@2 Requires bar-alias-template-requires.device
check_ok bar-alias@2 Requires yup-template-requires.device
check_ok bar-alias@2 Requires yup-2-requires.device
echo '*** bar-alias@3 aliases yup@3, bar@3 is independent ***'
systemctl show -p Names,Requires bar@3
systemctl show -p Names,Requires bar-alias@3
check_ok bar@3 Names bar@3
check_ko bar@3 Names bar-alias@3
check_ok bar@3 After bar-template-after.device
check_ok bar@3 Requires bar-3-requires.device
check_ko bar@3 Requires bar-alias-3-requires.device
check_ok bar@3 Requires bar-template-requires.device
check_ko bar@3 Requires bar-alias-template-requires.device
check_ko bar@3 Requires yup-template-requires.device
check_ko bar@3 Requires yup-3-requires.device
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
check_ko bar-alias@3 After bar-template-after.device
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
check_ko bar-alias@3 Requires bar-3-requires.device
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
check_ok bar-alias@3 Requires bar-alias-3-requires.device
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
check_ko bar-alias@3 Requires bar-template-requires.device
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
check_ok bar-alias@3 Requires bar-alias-template-requires.device
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
check_ok bar-alias@3 Requires yup-template-requires.device
check_ok bar-alias@3 Requires yup-3-requires.device
TEST-15-DROPIN: add test for details of unit aliasing I adjusted the tests to pass. I don't think the behaviour makes much sense, even if we ignore the issue with "lazy loading" of aliases. E.g. in the last section, the fact that dropins for yup@.service and yup@3.service are not loaded seems to be a plain old bug.
2019-07-30 10:28:48 +00:00
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units foo.service {bar,yup,bar-alias}@{,1,2,3}.service
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
}
test_alias_dropins () {
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
echo "Testing alias dropins..."
echo "*** test a wants b1 alias of b"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b
ln -sf test15-b.service /etc/systemd/system/test15-b1.service
ln -sf ../test15-b1.service /etc/systemd/system/test15-a.service.wants/
check_ok test15-a Wants test15-b.service
systemctl start test15-a
systemctl --quiet is-active test15-b
systemctl stop test15-a test15-b
rm /etc/systemd/system/test15-b1.service
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units test15-{a,b}.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
pid1: use a cache for all unit aliases This reworks how we load units from disk. Instead of chasing symlinks every time we are asked to load a unit by name, we slurp all symlinks from disk and build two hashmaps: 1. from unit name to either alias target, or fragment on disk (if an alias, we put just the target name in the hashmap, if a fragment we put an absolute path, so we can distinguish both). 2. from a unit name to all aliases Reading all this data can be pretty costly (40 ms) on my machine, so we keep it around for reuse. The advantage is that we can reliably know what all the aliases of a given unit are. This means we can reliably load dropins under all names. This fixes #11972.
2019-07-18 11:11:28 +00:00
# Check that dependencies don't vary.
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
echo "*** test 2"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-x test15-y
mkdir -p /etc/systemd/system/test15-a1.service.wants/
ln -sf test15-a.service /etc/systemd/system/test15-a1.service
ln -sf ../test15-x.service /etc/systemd/system/test15-a.service.wants/
ln -sf ../test15-y.service /etc/systemd/system/test15-a1.service.wants/
check_ok test15-a1 Wants test15-x.service # see [1]
check_ok test15-a1 Wants test15-y.service
systemctl start test15-a
check_ok test15-a1 Wants test15-x.service # see [2]
check_ok test15-a1 Wants test15-y.service
systemctl stop test15-a test15-x test15-y
rm /etc/systemd/system/test15-a1.service
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units test15-{a,x,y}.service
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
}
test_masked_dropins () {
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
echo "Testing masked dropins..."
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'b' is masked for both deps
echo "*** test a wants,requires b is masked"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
ln -sf /dev/null /etc/systemd/system/test15-a.service.wants/test15-b.service
ln -sf /dev/null /etc/systemd/system/test15-a.service.requires/test15-b.service
check_ko test15-a Wants test15-b.service
check_ko test15-a Requires test15-b.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'a' wants 'b' and 'b' is masked at a lower level
echo "*** test a wants b, mask override"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
ln -sf ../test15-b.service /etc/systemd/system/test15-a.service.wants/test15-b.service
ln -sf /dev/null /usr/lib/systemd/system/test15-a.service.wants/test15-b.service
check_ok test15-a Wants test15-b.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'a' wants 'b' and 'b' is masked at a higher level
echo "*** test a wants b, mask"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
ln -sf /dev/null /etc/systemd/system/test15-a.service.wants/test15-b.service
ln -sf ../test15-b.service /usr/lib/systemd/system/test15-a.service.wants/test15-b.service
check_ko test15-a Wants test15-b.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'a' is masked but has an override config file
echo "*** test a is masked but has an override"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b
ln -sf /dev/null /etc/systemd/system/test15-a.service
cat >/usr/lib/systemd/system/test15-a.service.d/override.conf <<EOF
test: masked unit with drop-ins
2018-03-02 05:07:27 +00:00
[Unit]
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
After=test15-b.service
test: masked unit with drop-ins
2018-03-02 05:07:27 +00:00
EOF
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
check_ok test15-a UnitFileState masked
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'b1' is an alias for 'b': masking 'b' dep should not influence 'b1' dep
echo "*** test a wants b, b1, and one is masked"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b
ln -sf test15-b.service /etc/systemd/system/test15-b1.service
ln -sf /dev/null /etc/systemd/system/test15-a.service.wants/test15-b.service
ln -sf ../test15-b1.service /usr/lib/systemd/system/test15-a.service.wants/test15-b1.service
systemctl cat test15-a
systemctl show -p Wants,Requires test15-a
systemctl cat test15-b1
systemctl show -p Wants,Requires test15-b1
check_ok test15-a Wants test15-b.service
check_ko test15-a Wants test15-b1.service # the alias does not show up in the list of units
rm /etc/systemd/system/test15-b1.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'b1' is an alias for 'b': masking 'b1' should not influence 'b' dep
echo "*** test a wants b, alias dep is masked"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b
ln -sf test15-b.service /etc/systemd/system/test15-b1.service
ln -sf /dev/null /etc/systemd/system/test15-a.service.wants/test15-b1.service
ln -sf ../test15-b.service /usr/lib/systemd/system/test15-a.service.wants/test15-b.service
check_ok test15-a Wants test15-b.service
check_ko test15-a Wants test15-b1.service # the alias does not show up in the list of units
rm /etc/systemd/system/test15-b1.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'a' has Wants=b.service but also has a masking
# dropin 'b': 'b' should still be pulled in.
echo "*** test a wants b both ways"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b
ln -sf /dev/null /etc/systemd/system/test15-a.service.wants/test15-b.service
cat >/usr/lib/systemd/system/test15-a.service.d/wants-b.conf<<EOF
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
[Unit]
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
Wants=test15-b.service
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
EOF
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
check_ok test15-a Wants test15-b.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# mask a dropin that points to an nonexistent unit.
echo "*** test a wants nonexistent is masked"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a
ln -sf /dev/null /etc/systemd/system/test15-a.service.requires/nonexistent.service
ln -sf ../nonexistent.service /usr/lib/systemd/system/test15-a.service.requires/
check_ko test15-a Requires nonexistent.service
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'b' is already loaded when 'c' pulls it in via a dropin but 'b' is
# masked at a higher level.
echo "*** test a wants b is masked"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b test15-c
ln -sf ../test15-b.service /etc/systemd/system/test15-a.service.requires/
ln -sf ../test15-b.service /run/systemd/system/test15-c.service.requires/
ln -sf /dev/null /etc/systemd/system/test15-c.service.requires/test15-b.service
systemctl start test15-a
check_ko test15-c Requires test15-b.service
systemctl stop test15-a test15-b
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'b' is already loaded when 'c' pulls it in via a dropin but 'b' is
# masked at a lower level.
echo "*** test a requires b is masked"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b test15-c
ln -sf ../test15-b.service /etc/systemd/system/test15-a.service.requires/
ln -sf ../test15-b.service /etc/systemd/system/test15-c.service.requires/
ln -sf /dev/null /run/systemd/system/test15-c.service.requires/test15-b.service
systemctl start test15-a
check_ok test15-c Requires test15-b.service
systemctl stop test15-a test15-b
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# 'a' requires 2 aliases of 'b' and one of them is a mask.
echo "*** test a requires alias of b, other alias masked"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b
ln -sf test15-b.service /etc/systemd/system/test15-b1.service
ln -sf test15-b.service /etc/systemd/system/test15-b2.service
ln -sf /dev/null /etc/systemd/system/test15-a.service.requires/test15-b1.service
ln -sf ../test15-b1.service /run/systemd/system/test15-a.service.requires/
ln -sf ../test15-b2.service /usr/lib/systemd/system/test15-a.service.requires/
check_ok test15-a Requires test15-b
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 12:10:42 +00:00
# Same as above but now 'b' is masked.
echo "*** test a requires alias of b, b dep masked"
test: move TEST-15-DROPIN setup to static files Unfortunately the services needed to be renamed because of conflicts with other services that exist in the shared dir.
2019-12-09 22:29:18 +00:00
create_services test15-a test15-b
ln -sf test15-b.service /etc/systemd/system/test15-b1.service
ln -sf test15-b.service /etc/systemd/system/test15-b2.service
ln -sf ../test15-b1.service /run/systemd/system/test15-a.service.requires/
ln -sf ../test15-b2.service /usr/lib/systemd/system/test15-a.service.requires/
ln -sf /dev/null /etc/systemd/system/test15-a.service.requires/test15-b.service
check_ok test15-a Requires test15-b
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units test15-{a,b}.service
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
}
shared/dropin: fix assert for invalid drop-in Don't try to show top level drop-in for non-existent units or when trying to instantiate non-instantiated units: $ systemctl cat nonexistent@.service Assertion 'name' failed at src/shared/dropin.c:143, function unit_file_find_dirs(). Aborting. $ systemctl cat systemd-journald@.service Assertion 'name' failed at src/shared/dropin.c:143, function unit_file_find_dirs(). Aborting.
2019-12-17 13:47:37 +00:00
test_invalid_dropins () {
echo "Testing invalid dropins..."
# Assertion failed on earlier versions, command exits unsuccessfully on later versions
systemctl cat nonexistent@.service || true
create_services a
systemctl daemon-reload
# Assertion failed on earlier versions, command exits unsuccessfully on later versions
systemctl cat a@.service || true
systemctl stop a
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units a.service
shared/dropin: fix assert for invalid drop-in Don't try to show top level drop-in for non-existent units or when trying to instantiate non-instantiated units: $ systemctl cat nonexistent@.service Assertion 'name' failed at src/shared/dropin.c:143, function unit_file_find_dirs(). Aborting. $ systemctl cat systemd-journald@.service Assertion 'name' failed at src/shared/dropin.c:143, function unit_file_find_dirs(). Aborting.
2019-12-17 13:47:37 +00:00
return 0
}
test: add testcases of symlinked drop-in directories
2021-12-28 15:49:11 +00:00
test_symlink_dropin_directory () {
# For issue #21920.
echo "Testing symlink drop-in directory..."
create_services test15-a
rmdir /{etc,run,usr/lib}/systemd/system/test15-a.service.d
mkdir -p /tmp/testsuite-15-test15-a-dropin-directory
ln -s /tmp/testsuite-15-test15-a-dropin-directory /etc/systemd/system/test15-a.service.d
cat >/tmp/testsuite-15-test15-a-dropin-directory/override.conf <<EOF
[Unit]
Description=hogehoge
EOF
ln -s /tmp/testsuite-15-test15-a-dropin-directory-nonexistent /run/systemd/system/test15-a.service.d
touch /tmp/testsuite-15-test15-a-dropin-directory-regular
ln -s /tmp/testsuite-15-test15-a-dropin-directory-regular /usr/lib/systemd/system/test15-a.service.d
check_ok test15-a Description hogehoge
TEST-15: allow helper functions to accept other unit types clear_services() is renamed to clear_units() and now takes a full unit name including the suffix as an argument. _clear_service() is renamed to clear_unit() and changed likewise. create_service() didn't have the same underscore prefix, and I don't think it's useful or needed for a local function, so it is removed. No functional change.
2022-10-16 10:42:35 +00:00
clear_units test15-a.service
test: add testcases of symlinked drop-in directories
2021-12-28 15:49:11 +00:00
}
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
test_basic_dropins
TEST-15-DROPINS: add test for linked units
2021-02-14 14:08:49 +00:00
test_linked_units
TEST-15: add test for template aliases
2021-04-29 07:47:23 +00:00
test_template_alias
TEST-15: also test hierarchical drop-ins for slices Slices are worth testing too, because they don't need a fragment path so they behave slightly differently than service units. I'm making this a separate patch from the actual tests that I wanted to add later because it's complex enough on its own.
2022-10-16 10:54:34 +00:00
test_hierarchical_service_dropins
test_hierarchical_slice_dropins
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
test_template_dropins
test_alias_dropins
test_masked_dropins
shared/dropin: fix assert for invalid drop-in Don't try to show top level drop-in for non-existent units or when trying to instantiate non-instantiated units: $ systemctl cat nonexistent@.service Assertion 'name' failed at src/shared/dropin.c:143, function unit_file_find_dirs(). Aborting. $ systemctl cat systemd-journald@.service Assertion 'name' failed at src/shared/dropin.c:143, function unit_file_find_dirs(). Aborting.
2019-12-17 13:47:37 +00:00
test_invalid_dropins
test: add testcases of symlinked drop-in directories
2021-12-28 15:49:11 +00:00
test_symlink_dropin_directory
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 15:02:22 +00:00
touch /testok
Reference in a new issue Copy permalink