shaba/systemd
1
0
Fork 0
mirror of https://github.com/systemd/systemd.git synced 2024-11-06 08:26:52 +03:00
Code
7a670b1dd9
systemd/test/TEST-15-DROPIN/test-dropin.sh

442 lines
16 KiB
Bash
Raw Normal View History

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 18:02:22 +03:00
#! /bin/bash
set -e
set -x
_clear_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 15:10:42 +03:00
systemctl stop $1.service 2>/dev/null || :
rm -f /{etc,run,usr/lib}/systemd/system/$1.service
rm -fr /{etc,run,usr/lib}/systemd/system/$1.service.d
rm -fr /{etc,run,usr/lib}/systemd/system/$1.service.{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 18:02:22 +03:00
}
clear_services () {
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 15:10:42 +03:00
for u in $*; do
_clear_service $u
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 18:02:22 +03:00
}
create_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 15:10:42 +03:00
clear_services $1
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 18:02:22 +03:00
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 15:10:42 +03:00
cat >/etc/systemd/system/$1.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 18:02:22 +03:00
[Unit]
Description=$1 unit
[Service]
ExecStart=/bin/sleep 100000
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 15:10:42 +03:00
mkdir -p /{etc,run,usr/lib}/systemd/system/$1.service.d
mkdir -p /etc/systemd/system/$1.service.{wants,requires}
mkdir -p /run/systemd/system/$1.service.{wants,requires}
mkdir -p /usr/lib/systemd/system/$1.service.{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 18:02:22 +03:00
}
create_services () {
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 15:10:42 +03:00
for u in $*; do
create_service $u
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 18:02:22 +03:00
}
check_ok () {
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 15:10:42 +03:00
[ $# -eq 3 ] || return
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 18:02:22 +03:00
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 15:10:42 +03:00
x="$(systemctl show --value -p $2 $1)"
case "$x" in
*$3*) return 0 ;;
*) return 1 ;;
esac
tests: add dropin dependency tests [zj: tests assertions adjusted to the different logic in which masking of a dependency through one name, does not forbid the dependency being added through another name.]
2017-01-27 18:02:22 +03:00
}
check_ko () {
scripts: use 4 space indentation We had all kinds of indentation: 2 sp, 3 sp, 4 sp, 8 sp, and mixed. 4 sp was the most common, in particular the majority of scripts under test/ used that. Let's standarize on 4 sp, because many commandlines are long and there's a lot of nesting, and with 8sp indentation less stuff fits. 4 sp also seems to be the default indentation, so this will make it less likely that people will mess up if they don't load the editor config. (I think people often use vi, and vi has no support to load project-wide configuration automatically. We distribute a .vimrc file, but it is not loaded by default, and even the instructions in it seem to discourage its use for security reasons.) Also remove the few vim config lines that were left. We should either have them on all files, or none. Also remove some strange stuff like '#!/bin/env bash', yikes.
2019-04-04 15:10:42 +03:00
! check_ok "$@"
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 18:02:22 +03:00
}
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 15:10:42 +03:00
echo "Testing basic dropins..."
echo "*** test a wants b wants c"
create_services a b c
ln -s ../b.service /etc/systemd/system/a.service.wants/
ln -s ../c.service /etc/systemd/system/b.service.wants/
check_ok a Wants b.service
check_ok b Wants c.service
echo "*** test a wants,requires b"
create_services a b c
ln -s ../b.service /etc/systemd/system/a.service.wants/
ln -s ../b.service /etc/systemd/system/a.service.requires/
check_ok a Wants b.service
check_ok a Requires b.service
echo "*** test a wants nonexistent"
create_service a
ln -s ../nonexistent.service /etc/systemd/system/a.service.wants/
check_ok a Wants nonexistent.service
systemctl start a
systemctl stop a
echo "*** test a requires nonexistent"
ln -sf ../nonexistent.service /etc/systemd/system/a.service.requires/
systemctl daemon-reload
check_ok a Requires nonexistent.service
# 'b' is already loaded when 'c' pulls it in via a dropin.
echo "*** test a,c require b"
create_services a b c
ln -sf ../b.service /etc/systemd/system/a.service.requires/
ln -sf ../b.service /etc/systemd/system/c.service.requires/
systemctl start a
check_ok c Requires b.service
systemctl stop a b
# 'b' is already loaded when 'c' pulls it in via an alias dropin.
echo "*** test a wants alias"
create_services a b c
ln -sf c.service /etc/systemd/system/c1.service
ln -sf ../c.service /etc/systemd/system/a.service.wants/
ln -sf ../c1.service /etc/systemd/system/b.service.wants/
systemctl start a
check_ok a Wants c.service
check_ok b Wants c.service
systemctl stop a c
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 05:29:55 +03:00
echo "*** test service.d/ top level drop-in"
shared/dropin: support -.service.d/ top level drop-in for service units Closes #12830
2019-10-05 03:39:34 +03:00
create_services a b
check_ko a ExecCondition "/bin/echo a"
check_ko 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 05:29:55 +03:00
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 03:39:34 +03:00
[Service]
ExecCondition=/bin/echo %n
EOF
check_ok a ExecCondition "/bin/echo a"
check_ok 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 05:29:55 +03:00
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 03:39:34 +03:00
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 15:10:42 +03:00
clear_services a b c
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 18:02:22 +03:00
}
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 15:10:42 +03:00
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 18:02:22 +03:00
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 15:10:42 +03:00
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 18:02:22 +03:00
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 13:28:48 +03:00
# 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 15:10:42 +03:00
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 18:02:22 +03:00
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 13:28:48 +03:00
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 14:11:28 +03:00
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 13:28:48 +03:00
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 14:11:28 +03:00
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 13:28:48 +03:00
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 14:11:28 +03:00
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 13:28:48 +03:00
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 14:11:28 +03:00
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 13:28:48 +03:00
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 14:11:28 +03:00
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 13:28:48 +03:00
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 14:11:28 +03:00
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 13:28:48 +03:00
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 14:11:28 +03:00
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 13:28:48 +03:00
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 14:11:28 +03:00
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 13:28:48 +03:00
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 14:11:28 +03:00
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 13:28:48 +03:00
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 14:11:28 +03:00
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 13:28:48 +03:00
clear_services foo {bar,yup,bar-alias}@{,1,2,3}
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 18:02:22 +03:00
}
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 15:10:42 +03:00
echo "Testing alias dropins..."
echo "*** test a wants b1 alias of b"
create_services a b
ln -sf b.service /etc/systemd/system/b1.service
ln -sf ../b1.service /etc/systemd/system/a.service.wants/
check_ok a Wants b.service
systemctl start a
systemctl --quiet is-active b
systemctl stop a b
rm /etc/systemd/system/b1.service
clear_services a b
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 14:11:28 +03:00
# 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 15:10:42 +03:00
echo "*** test 2"
create_services a x y
mkdir -p /etc/systemd/system/a1.service.wants/
ln -sf a.service /etc/systemd/system/a1.service
ln -sf ../x.service /etc/systemd/system/a.service.wants/
ln -sf ../y.service /etc/systemd/system/a1.service.wants/
check_ok a1 Wants x.service # see [1]
check_ok a1 Wants y.service
systemctl start a
check_ok a1 Wants x.service # see [2]
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 14:11:28 +03:00
check_ok a1 Wants y.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 15:10:42 +03:00
systemctl stop a x y
rm /etc/systemd/system/a1.service
clear_services a x y
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 18:02:22 +03:00
}
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 15:10:42 +03:00
echo "Testing masked dropins..."
create_services a b
# 'b' is masked for both deps
echo "*** test a wants,requires b is masked"
ln -sf /dev/null /etc/systemd/system/a.service.wants/b.service
ln -sf /dev/null /etc/systemd/system/a.service.requires/b.service
check_ko a Wants b.service
check_ko a Requires b.service
# 'a' wants 'b' and 'b' is masked at a lower level
echo "*** test a wants b, mask override"
ln -sf ../b.service /etc/systemd/system/a.service.wants/b.service
ln -sf /dev/null /usr/lib/systemd/system/a.service.wants/b.service
check_ok a Wants b.service
# 'a' wants 'b' and 'b' is masked at a higher level
echo "*** test a wants b, mask"
ln -sf /dev/null /etc/systemd/system/a.service.wants/b.service
ln -sf ../b.service /usr/lib/systemd/system/a.service.wants/b.service
check_ko a Wants b.service
# 'a' is masked but has an override config file
echo "*** test a is masked but has an override"
create_services a b
ln -sf /dev/null /etc/systemd/system/a.service
cat >/usr/lib/systemd/system/a.service.d/override.conf <<EOF
test: masked unit with drop-ins
2018-03-02 08:07:27 +03:00
[Unit]
After=b.service
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 15:10:42 +03:00
check_ok a UnitFileState masked
# 'b1' is an alias for 'b': masking 'b' dep should not influence 'b1' dep
echo "*** test a wants b, b1, and one is masked"
create_services a b
ln -sf b.service /etc/systemd/system/b1.service
ln -sf /dev/null /etc/systemd/system/a.service.wants/b.service
ln -sf ../b1.service /usr/lib/systemd/system/a.service.wants/b1.service
systemctl cat a
systemctl show -p Wants,Requires a
systemctl cat b1
systemctl show -p Wants,Requires b1
check_ok a Wants b.service
check_ko a Wants b1.service # the alias does not show up in the list of units
rm /etc/systemd/system/b1.service
# 'b1' is an alias for 'b': masking 'b1' should not influence 'b' dep
echo "*** test a wants b, alias dep is masked"
create_services a b
ln -sf b.service /etc/systemd/system/b1.service
ln -sf /dev/null /etc/systemd/system/a.service.wants/b1.service
ln -sf ../b.service /usr/lib/systemd/system/a.service.wants/b.service
check_ok a Wants b.service
check_ko a Wants b1.service # the alias does not show up in the list of units
rm /etc/systemd/system/b1.service
# '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"
create_services a b
ln -sf /dev/null /etc/systemd/system/a.service.wants/b.service
cat >/usr/lib/systemd/system/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 18:02:22 +03:00
[Unit]
Wants=b.service
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 15:10:42 +03:00
check_ok a Wants b.service
# mask a dropin that points to an nonexistent unit.
echo "*** test a wants nonexistent is masked"
create_services a
ln -sf /dev/null /etc/systemd/system/a.service.requires/nonexistent.service
ln -sf ../nonexistent.service /usr/lib/systemd/system/a.service.requires/
check_ko a Requires nonexistent.service
# '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"
create_services a b c
ln -sf ../b.service /etc/systemd/system/a.service.requires/
ln -sf ../b.service /run/systemd/system/c.service.requires/
ln -sf /dev/null /etc/systemd/system/c.service.requires/b.service
systemctl start a
check_ko c Requires b.service
systemctl stop a b
# '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"
create_services a b c
ln -sf ../b.service /etc/systemd/system/a.service.requires/
ln -sf ../b.service /etc/systemd/system/c.service.requires/
ln -sf /dev/null /run/systemd/system/c.service.requires/b.service
systemctl start a
check_ok c Requires b.service
systemctl stop a b
# 'a' requires 2 aliases of 'b' and one of them is a mask.
echo "*** test a requires alias of b, other alias masked"
create_services a b
ln -sf b.service /etc/systemd/system/b1.service
ln -sf b.service /etc/systemd/system/b2.service
ln -sf /dev/null /etc/systemd/system/a.service.requires/b1.service
ln -sf ../b1.service /run/systemd/system/a.service.requires/
ln -sf ../b2.service /usr/lib/systemd/system/a.service.requires/
check_ok a Requires b
# Same as above but now 'b' is masked.
echo "*** test a requires alias of b, b dep masked"
create_services a b
ln -sf b.service /etc/systemd/system/b1.service
ln -sf b.service /etc/systemd/system/b2.service
ln -sf ../b1.service /run/systemd/system/a.service.requires/
ln -sf ../b2.service /usr/lib/systemd/system/a.service.requires/
ln -sf /dev/null /etc/systemd/system/a.service.requires/b.service
check_ok a Requires b
clear_services a b
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 18:02:22 +03:00
}
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 16:47:37 +03:00
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
clear_services a
return 0
}
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 18:02:22 +03:00
test_basic_dropins
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 16:47:37 +03:00
test_invalid_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 18:02:22 +03:00
touch /testok
Copy Permalink