moby/daemon/daemon_linux_test.go

399 lines
19 KiB
Go
Raw Permalink Normal View History

//go:build linux
package daemon // import "github.com/docker/docker/daemon"
import (
2023-01-16 22:25:13 +00:00
"net"
"os"
"path/filepath"
"strings"
"testing"
containertypes "github.com/docker/docker/api/types/container"
"github.com/docker/docker/internal/testutils/netnsutils"
2023-01-16 22:25:13 +00:00
"github.com/docker/docker/libnetwork/types"
"github.com/google/go-cmp/cmp/cmpopts"
"github.com/moby/sys/mount"
"github.com/moby/sys/mountinfo"
2023-01-16 22:25:13 +00:00
"github.com/vishvananda/netlink"
"gotest.tools/v3/assert"
is "gotest.tools/v3/assert/cmp"
)
//nolint:dupword
const mountsFixture = `142 78 0:38 / / rw,relatime - aufs none rw,si=573b861da0b3a05b,dio
143 142 0:60 / /proc rw,nosuid,nodev,noexec,relatime - proc proc rw
144 142 0:67 / /dev rw,nosuid - tmpfs tmpfs rw,mode=755
145 144 0:78 / /dev/pts rw,nosuid,noexec,relatime - devpts devpts rw,gid=5,mode=620,ptmxmode=666
146 144 0:49 / /dev/mqueue rw,nosuid,nodev,noexec,relatime - mqueue mqueue rw
147 142 0:84 / /sys rw,nosuid,nodev,noexec,relatime - sysfs sysfs rw
148 147 0:86 / /sys/fs/cgroup rw,nosuid,nodev,noexec,relatime - tmpfs tmpfs rw,mode=755
149 148 0:22 /docker/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a /sys/fs/cgroup/cpuset rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,cpuset
150 148 0:25 /docker/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a /sys/fs/cgroup/cpu rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,cpu
151 148 0:27 /docker/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a /sys/fs/cgroup/cpuacct rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,cpuacct
152 148 0:28 /docker/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a /sys/fs/cgroup/memory rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,memory
153 148 0:29 /docker/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a /sys/fs/cgroup/devices rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,devices
154 148 0:30 /docker/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a /sys/fs/cgroup/freezer rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,freezer
155 148 0:31 /docker/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a /sys/fs/cgroup/blkio rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,blkio
156 148 0:32 /docker/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a /sys/fs/cgroup/perf_event rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,perf_event
157 148 0:33 /docker/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a /sys/fs/cgroup/hugetlb rw,nosuid,nodev,noexec,relatime - cgroup cgroup rw,hugetlb
158 148 0:35 /docker/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a /sys/fs/cgroup/systemd rw,nosuid,nodev,noexec,relatime - cgroup systemd rw,name=systemd
159 142 8:4 /home/mlaventure/gopath /home/mlaventure/gopath rw,relatime - ext4 /dev/disk/by-uuid/d99e196c-1fc4-4b4f-bab9-9962b2b34e99 rw,errors=remount-ro,data=ordered
160 142 8:4 /var/lib/docker/volumes/9a428b651ee4c538130143cad8d87f603a4bf31b928afe7ff3ecd65480692b35/_data /var/lib/docker rw,relatime - ext4 /dev/disk/by-uuid/d99e196c-1fc4-4b4f-bab9-9962b2b34e99 rw,errors=remount-ro,data=ordered
164 142 8:4 /home/mlaventure/gopath/src/github.com/docker/docker /go/src/github.com/docker/docker rw,relatime - ext4 /dev/disk/by-uuid/d99e196c-1fc4-4b4f-bab9-9962b2b34e99 rw,errors=remount-ro,data=ordered
165 142 8:4 /var/lib/docker/containers/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a/resolv.conf /etc/resolv.conf rw,relatime - ext4 /dev/disk/by-uuid/d99e196c-1fc4-4b4f-bab9-9962b2b34e99 rw,errors=remount-ro,data=ordered
166 142 8:4 /var/lib/docker/containers/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a/hostname /etc/hostname rw,relatime - ext4 /dev/disk/by-uuid/d99e196c-1fc4-4b4f-bab9-9962b2b34e99 rw,errors=remount-ro,data=ordered
167 142 8:4 /var/lib/docker/containers/5425782a95e643181d8a485a2bab3c0bb21f51d7dfc03511f0e6fbf3f3aa356a/hosts /etc/hosts rw,relatime - ext4 /dev/disk/by-uuid/d99e196c-1fc4-4b4f-bab9-9962b2b34e99 rw,errors=remount-ro,data=ordered
168 144 0:39 / /dev/shm rw,nosuid,nodev,noexec,relatime - tmpfs shm rw,size=65536k
169 144 0:12 /14 /dev/console rw,nosuid,noexec,relatime - devpts devpts rw,gid=5,mode=620,ptmxmode=000
83 147 0:10 / /sys/kernel/security rw,relatime - securityfs none rw
89 142 0:87 / /tmp rw,relatime - tmpfs none rw
97 142 0:60 / /run/docker/netns/default rw,nosuid,nodev,noexec,relatime - proc proc rw
100 160 8:4 /var/lib/docker/volumes/9a428b651ee4c538130143cad8d87f603a4bf31b928afe7ff3ecd65480692b35/_data/aufs /var/lib/docker/aufs rw,relatime - ext4 /dev/disk/by-uuid/d99e196c-1fc4-4b4f-bab9-9962b2b34e99 rw,errors=remount-ro,data=ordered
115 100 0:102 / /var/lib/docker/aufs/mnt/0ecda1c63e5b58b3d89ff380bf646c95cc980252cf0b52466d43619aec7c8432 rw,relatime - aufs none rw,si=573b861dbc01905b,dio
116 160 0:107 / /var/lib/docker/containers/d045dc441d2e2e1d5b3e328d47e5943811a40819fb47497c5f5a5df2d6d13c37/shm rw,nosuid,nodev,noexec,relatime - tmpfs shm rw,size=65536k
118 142 0:102 / /run/docker/libcontainerd/d045dc441d2e2e1d5b3e328d47e5943811a40819fb47497c5f5a5df2d6d13c37/rootfs rw,relatime - aufs none rw,si=573b861dbc01905b,dio
242 142 0:60 / /run/docker/netns/c3664df2a0f7 rw,nosuid,nodev,noexec,relatime - proc proc rw
120 100 0:122 / /var/lib/docker/aufs/mnt/03ca4b49e71f1e49a41108829f4d5c70ac95934526e2af8984a1f65f1de0715d rw,relatime - aufs none rw,si=573b861eb147805b,dio
171 142 0:122 / /run/docker/libcontainerd/e406ff6f3e18516d50e03dbca4de54767a69a403a6f7ec1edc2762812824521e/rootfs rw,relatime - aufs none rw,si=573b861eb147805b,dio
310 142 0:60 / /run/docker/netns/71a18572176b rw,nosuid,nodev,noexec,relatime - proc proc rw
`
//nolint:dupword
daemon: fix daemon.Shutdown, daemon.Cleanup not cleaning up overlay2 mounts While working on deprecation of the `aufs` and `overlay` storage-drivers, the `TestCleanupMounts` had to be updated, as it was currently using `aufs` for testing. When rewriting the test to use `overlay2` instead (using an updated `mountsFixture`), I found out that the test was failing, and it appears that only `overlay`, but not `overlay2` was taken into account. These cleanup functions were added in 05cc737f5411a0effd299429140d031c4ad8dd05, but at the time the `overlay2` storage driver was not yet implemented; https://github.com/moby/moby/tree/05cc737f5411a0effd299429140d031c4ad8dd05/daemon/graphdriver This omission was likely missed in 23e5c94cfb26eb72c097892712d3dbaa93ee9bc0, because the original implementation re-used the `overlay` storage driver, but later on it was decided to make `overlay2` a separate storage driver. As a result of the above, `daemon.cleanupMountsByID()` would ignore any `overlay2` mounts during `daemon.Shutdown()` and `daemon.Cleanup()`. This patch: - Adds a new `mountsFixtureOverlay2` with example mounts for `overlay2` - Rewrites the tests to use `gotest.tools` for more informative output on failures. - Adds the missing regex patterns to `daemon/getCleanPatterns()`. The patterns are added at the start of the list to allow for the fasted match (`overlay2` is the default for most setups, and the code is iterating over possible options). As a follow-up, we could consider adding additional fixtures for different storage drivers. Before the fix is applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 daemon_linux_test.go:135: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the shm (and the shm only) --- FAIL: TestCleanupMounts (0.01s) --- PASS: TestCleanupMounts/aufs (0.00s) --- FAIL: TestCleanupMounts/overlay2 (0.01s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 daemon_linux_test.go:171: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the root (and that only) --- FAIL: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- FAIL: TestCleanupMountsByID/overlay2 (0.00s) FAIL FAIL github.com/docker/docker/daemon 0.054s FAIL With the fix applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 --- PASS: TestCleanupMounts (0.00s) --- PASS: TestCleanupMounts/aufs (0.00s) --- PASS: TestCleanupMounts/overlay2 (0.00s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 --- PASS: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- PASS: TestCleanupMountsByID/overlay2 (0.00s) PASS ok github.com/docker/docker/daemon 0.042s Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2022-05-29 14:20:14 +00:00
const mountsFixtureOverlay2 = `23 28 0:22 / /sys rw,nosuid,nodev,noexec,relatime shared:7 - sysfs sysfs rw
24 28 0:4 / /proc rw,nosuid,nodev,noexec,relatime shared:13 - proc proc rw
25 28 0:6 / /dev rw,nosuid,relatime shared:2 - devtmpfs udev rw,size=491380k,nr_inodes=122845,mode=755
26 25 0:23 / /dev/pts rw,nosuid,noexec,relatime shared:3 - devpts devpts rw,gid=5,mode=620,ptmxmode=000
27 28 0:24 / /run rw,nosuid,noexec,relatime shared:5 - tmpfs tmpfs rw,size=100884k,mode=755
28 0 252:1 / / rw,relatime shared:1 - ext4 /dev/vda1 rw,data=ordered
29 23 0:7 / /sys/kernel/security rw,nosuid,nodev,noexec,relatime shared:8 - securityfs securityfs rw
30 25 0:25 / /dev/shm rw,nosuid,nodev shared:4 - tmpfs tmpfs rw
31 27 0:26 / /run/lock rw,nosuid,nodev,noexec,relatime shared:6 - tmpfs tmpfs rw,size=5120k
32 23 0:27 / /sys/fs/cgroup ro,nosuid,nodev,noexec shared:9 - tmpfs tmpfs ro,mode=755
33 32 0:28 / /sys/fs/cgroup/unified rw,nosuid,nodev,noexec,relatime shared:10 - cgroup2 cgroup rw
34 32 0:29 / /sys/fs/cgroup/systemd rw,nosuid,nodev,noexec,relatime shared:11 - cgroup cgroup rw,xattr,name=systemd
35 23 0:30 / /sys/fs/pstore rw,nosuid,nodev,noexec,relatime shared:12 - pstore pstore rw
36 32 0:31 / /sys/fs/cgroup/blkio rw,nosuid,nodev,noexec,relatime shared:14 - cgroup cgroup rw,blkio
37 32 0:32 / /sys/fs/cgroup/memory rw,nosuid,nodev,noexec,relatime shared:15 - cgroup cgroup rw,memory
38 32 0:33 / /sys/fs/cgroup/hugetlb rw,nosuid,nodev,noexec,relatime shared:16 - cgroup cgroup rw,hugetlb
39 32 0:34 / /sys/fs/cgroup/freezer rw,nosuid,nodev,noexec,relatime shared:17 - cgroup cgroup rw,freezer
40 32 0:35 / /sys/fs/cgroup/perf_event rw,nosuid,nodev,noexec,relatime shared:18 - cgroup cgroup rw,perf_event
41 32 0:36 / /sys/fs/cgroup/pids rw,nosuid,nodev,noexec,relatime shared:19 - cgroup cgroup rw,pids
42 32 0:37 / /sys/fs/cgroup/cpuset rw,nosuid,nodev,noexec,relatime shared:20 - cgroup cgroup rw,cpuset
43 32 0:38 / /sys/fs/cgroup/cpu,cpuacct rw,nosuid,nodev,noexec,relatime shared:21 - cgroup cgroup rw,cpu,cpuacct
44 32 0:39 / /sys/fs/cgroup/rdma rw,nosuid,nodev,noexec,relatime shared:22 - cgroup cgroup rw,rdma
45 32 0:40 / /sys/fs/cgroup/devices rw,nosuid,nodev,noexec,relatime shared:23 - cgroup cgroup rw,devices
46 32 0:41 / /sys/fs/cgroup/net_cls,net_prio rw,nosuid,nodev,noexec,relatime shared:24 - cgroup cgroup rw,net_cls,net_prio
47 24 0:42 / /proc/sys/fs/binfmt_misc rw,relatime shared:25 - autofs systemd-1 rw,fd=33,pgrp=1,timeout=0,minproto=5,maxproto=5,direct,pipe_ino=11725
48 23 0:8 / /sys/kernel/debug rw,relatime shared:26 - debugfs debugfs rw
49 25 0:19 / /dev/mqueue rw,relatime shared:27 - mqueue mqueue rw
50 25 0:43 / /dev/hugepages rw,relatime shared:28 - hugetlbfs hugetlbfs rw,pagesize=2M
80 23 0:20 / /sys/kernel/config rw,relatime shared:29 - configfs configfs rw
82 23 0:44 / /sys/fs/fuse/connections rw,relatime shared:30 - fusectl fusectl rw
84 28 252:15 / /boot/efi rw,relatime shared:31 - vfat /dev/vda15 rw,fmask=0022,dmask=0022,codepage=437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro
391 28 0:49 / /var/lib/lxcfs rw,nosuid,nodev,relatime shared:208 - fuse.lxcfs lxcfs rw,user_id=0,group_id=0,allow_other
401 48 0:11 / /sys/kernel/debug/tracing rw,relatime shared:213 - tracefs tracefs rw
421 47 0:93 / /proc/sys/fs/binfmt_misc rw,relatime shared:223 - binfmt_misc binfmt_misc rw
510 27 0:3 net:[4026531993] /run/docker/netns/default rw shared:255 - nsfs nsfs rw
60 27 0:3 net:[4026532265] /run/docker/netns/ingress_sbox rw shared:40 - nsfs nsfs rw
162 27 0:3 net:[4026532331] /run/docker/netns/1-bj0aarwy1n rw shared:41 - nsfs nsfs rw
450 28 0:51 / /var/lib/docker/overlay2/3a4b807fcb98c208573f368c5654a6568545a7f92404a07d0045eb5c85acaf67/merged rw,relatime shared:231 - overlay overlay rw,lowerdir=/var/lib/docker/overlay2/l/E6KNVZ2QUCIXY5VT7E5LO3PVCA:/var/lib/docker/overlay2/l/64XI57TRGG6QS4K6DCSREZXBN2:/var/lib/docker/overlay2/l/TWXZ4ANJR6BDLDZMWZ4Y6AICAR:/var/lib/docker/overlay2/l/VRLSNSG3PKZELC5O66TVTQ7EH5:/var/lib/docker/overlay2/l/HOLV4F57X56TRLVACMRLFVW7YD:/var/lib/docker/overlay2/l/JJQFBBBT6LWLQS35XBADV6BLAM:/var/lib/docker/overlay2/l/FZTPKHZGP2Z6DBPFEEL2IK3I5Y,upperdir=/var/lib/docker/overlay2/3a4b807fcb98c208573f368c5654a6568545a7f92404a07d0045eb5c85acaf67/diff,workdir=/var/lib/docker/overlay2/3a4b807fcb98c208573f368c5654a6568545a7f92404a07d0045eb5c85acaf67/work
569 27 0:3 net:[4026532353] /run/docker/netns/7de1071d0d8b rw shared:245 - nsfs nsfs rw
245 27 0:50 / /run/user/0 rw,nosuid,nodev,relatime shared:160 - tmpfs tmpfs rw,size=100880k,mode=700
482 28 0:69 / /var/lib/docker/overlay2/df4ee7b0bac7bda30e6e3d24a1153b288ebda50ffe68aae7ae0f38bc9286a01a/merged rw,relatime shared:250 - overlay overlay rw,lowerdir=/var/lib/docker/overlay2/l/CNZ3ATGGHMUTPPJBBU2OL4GLL6:/var/lib/docker/overlay2/l/64XI57TRGG6QS4K6DCSREZXBN2:/var/lib/docker/overlay2/l/TWXZ4ANJR6BDLDZMWZ4Y6AICAR:/var/lib/docker/overlay2/l/VRLSNSG3PKZELC5O66TVTQ7EH5:/var/lib/docker/overlay2/l/HOLV4F57X56TRLVACMRLFVW7YD:/var/lib/docker/overlay2/l/JJQFBBBT6LWLQS35XBADV6BLAM:/var/lib/docker/overlay2/l/FZTPKHZGP2Z6DBPFEEL2IK3I5Y,upperdir=/var/lib/docker/overlay2/df4ee7b0bac7bda30e6e3d24a1153b288ebda50ffe68aae7ae0f38bc9286a01a/diff,workdir=/var/lib/docker/overlay2/df4ee7b0bac7bda30e6e3d24a1153b288ebda50ffe68aae7ae0f38bc9286a01a/work
528 28 0:77 / /var/lib/docker/containers/404a7f860e600bfc144f7b5d9140d80bf3072fbb97659f98bc47039fd73d2695/mounts/shm rw,nosuid,nodev,noexec,relatime shared:260 - tmpfs shm rw,size=65536k
649 27 0:3 net:[4026532429] /run/docker/netns/7f85bc5ef3ba rw shared:265 - nsfs nsfs rw
`
func TestCleanupMounts(t *testing.T) {
d := &Daemon{
root: "/var/lib/docker/",
}
daemon: fix daemon.Shutdown, daemon.Cleanup not cleaning up overlay2 mounts While working on deprecation of the `aufs` and `overlay` storage-drivers, the `TestCleanupMounts` had to be updated, as it was currently using `aufs` for testing. When rewriting the test to use `overlay2` instead (using an updated `mountsFixture`), I found out that the test was failing, and it appears that only `overlay`, but not `overlay2` was taken into account. These cleanup functions were added in 05cc737f5411a0effd299429140d031c4ad8dd05, but at the time the `overlay2` storage driver was not yet implemented; https://github.com/moby/moby/tree/05cc737f5411a0effd299429140d031c4ad8dd05/daemon/graphdriver This omission was likely missed in 23e5c94cfb26eb72c097892712d3dbaa93ee9bc0, because the original implementation re-used the `overlay` storage driver, but later on it was decided to make `overlay2` a separate storage driver. As a result of the above, `daemon.cleanupMountsByID()` would ignore any `overlay2` mounts during `daemon.Shutdown()` and `daemon.Cleanup()`. This patch: - Adds a new `mountsFixtureOverlay2` with example mounts for `overlay2` - Rewrites the tests to use `gotest.tools` for more informative output on failures. - Adds the missing regex patterns to `daemon/getCleanPatterns()`. The patterns are added at the start of the list to allow for the fasted match (`overlay2` is the default for most setups, and the code is iterating over possible options). As a follow-up, we could consider adding additional fixtures for different storage drivers. Before the fix is applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 daemon_linux_test.go:135: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the shm (and the shm only) --- FAIL: TestCleanupMounts (0.01s) --- PASS: TestCleanupMounts/aufs (0.00s) --- FAIL: TestCleanupMounts/overlay2 (0.01s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 daemon_linux_test.go:171: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the root (and that only) --- FAIL: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- FAIL: TestCleanupMountsByID/overlay2 (0.00s) FAIL FAIL github.com/docker/docker/daemon 0.054s FAIL With the fix applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 --- PASS: TestCleanupMounts (0.00s) --- PASS: TestCleanupMounts/aufs (0.00s) --- PASS: TestCleanupMounts/overlay2 (0.00s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 --- PASS: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- PASS: TestCleanupMountsByID/overlay2 (0.00s) PASS ok github.com/docker/docker/daemon 0.042s Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2022-05-29 14:20:14 +00:00
t.Run("aufs", func(t *testing.T) {
expected := "/var/lib/docker/containers/d045dc441d2e2e1d5b3e328d47e5943811a40819fb47497c5f5a5df2d6d13c37/shm"
var unmounted int
unmount := func(target string) error {
if target == expected {
unmounted++
}
return nil
}
daemon: fix daemon.Shutdown, daemon.Cleanup not cleaning up overlay2 mounts While working on deprecation of the `aufs` and `overlay` storage-drivers, the `TestCleanupMounts` had to be updated, as it was currently using `aufs` for testing. When rewriting the test to use `overlay2` instead (using an updated `mountsFixture`), I found out that the test was failing, and it appears that only `overlay`, but not `overlay2` was taken into account. These cleanup functions were added in 05cc737f5411a0effd299429140d031c4ad8dd05, but at the time the `overlay2` storage driver was not yet implemented; https://github.com/moby/moby/tree/05cc737f5411a0effd299429140d031c4ad8dd05/daemon/graphdriver This omission was likely missed in 23e5c94cfb26eb72c097892712d3dbaa93ee9bc0, because the original implementation re-used the `overlay` storage driver, but later on it was decided to make `overlay2` a separate storage driver. As a result of the above, `daemon.cleanupMountsByID()` would ignore any `overlay2` mounts during `daemon.Shutdown()` and `daemon.Cleanup()`. This patch: - Adds a new `mountsFixtureOverlay2` with example mounts for `overlay2` - Rewrites the tests to use `gotest.tools` for more informative output on failures. - Adds the missing regex patterns to `daemon/getCleanPatterns()`. The patterns are added at the start of the list to allow for the fasted match (`overlay2` is the default for most setups, and the code is iterating over possible options). As a follow-up, we could consider adding additional fixtures for different storage drivers. Before the fix is applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 daemon_linux_test.go:135: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the shm (and the shm only) --- FAIL: TestCleanupMounts (0.01s) --- PASS: TestCleanupMounts/aufs (0.00s) --- FAIL: TestCleanupMounts/overlay2 (0.01s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 daemon_linux_test.go:171: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the root (and that only) --- FAIL: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- FAIL: TestCleanupMountsByID/overlay2 (0.00s) FAIL FAIL github.com/docker/docker/daemon 0.054s FAIL With the fix applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 --- PASS: TestCleanupMounts (0.00s) --- PASS: TestCleanupMounts/aufs (0.00s) --- PASS: TestCleanupMounts/overlay2 (0.00s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 --- PASS: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- PASS: TestCleanupMountsByID/overlay2 (0.00s) PASS ok github.com/docker/docker/daemon 0.042s Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2022-05-29 14:20:14 +00:00
err := d.cleanupMountsFromReaderByID(strings.NewReader(mountsFixture), "", unmount)
assert.NilError(t, err)
assert.Equal(t, unmounted, 1, "Expected to unmount the shm (and the shm only)")
})
daemon: fix daemon.Shutdown, daemon.Cleanup not cleaning up overlay2 mounts While working on deprecation of the `aufs` and `overlay` storage-drivers, the `TestCleanupMounts` had to be updated, as it was currently using `aufs` for testing. When rewriting the test to use `overlay2` instead (using an updated `mountsFixture`), I found out that the test was failing, and it appears that only `overlay`, but not `overlay2` was taken into account. These cleanup functions were added in 05cc737f5411a0effd299429140d031c4ad8dd05, but at the time the `overlay2` storage driver was not yet implemented; https://github.com/moby/moby/tree/05cc737f5411a0effd299429140d031c4ad8dd05/daemon/graphdriver This omission was likely missed in 23e5c94cfb26eb72c097892712d3dbaa93ee9bc0, because the original implementation re-used the `overlay` storage driver, but later on it was decided to make `overlay2` a separate storage driver. As a result of the above, `daemon.cleanupMountsByID()` would ignore any `overlay2` mounts during `daemon.Shutdown()` and `daemon.Cleanup()`. This patch: - Adds a new `mountsFixtureOverlay2` with example mounts for `overlay2` - Rewrites the tests to use `gotest.tools` for more informative output on failures. - Adds the missing regex patterns to `daemon/getCleanPatterns()`. The patterns are added at the start of the list to allow for the fasted match (`overlay2` is the default for most setups, and the code is iterating over possible options). As a follow-up, we could consider adding additional fixtures for different storage drivers. Before the fix is applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 daemon_linux_test.go:135: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the shm (and the shm only) --- FAIL: TestCleanupMounts (0.01s) --- PASS: TestCleanupMounts/aufs (0.00s) --- FAIL: TestCleanupMounts/overlay2 (0.01s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 daemon_linux_test.go:171: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the root (and that only) --- FAIL: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- FAIL: TestCleanupMountsByID/overlay2 (0.00s) FAIL FAIL github.com/docker/docker/daemon 0.054s FAIL With the fix applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 --- PASS: TestCleanupMounts (0.00s) --- PASS: TestCleanupMounts/aufs (0.00s) --- PASS: TestCleanupMounts/overlay2 (0.00s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 --- PASS: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- PASS: TestCleanupMountsByID/overlay2 (0.00s) PASS ok github.com/docker/docker/daemon 0.042s Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2022-05-29 14:20:14 +00:00
t.Run("overlay2", func(t *testing.T) {
expected := "/var/lib/docker/containers/404a7f860e600bfc144f7b5d9140d80bf3072fbb97659f98bc47039fd73d2695/mounts/shm"
var unmounted int
unmount := func(target string) error {
if target == expected {
unmounted++
}
return nil
}
err := d.cleanupMountsFromReaderByID(strings.NewReader(mountsFixtureOverlay2), "", unmount)
assert.NilError(t, err)
assert.Equal(t, unmounted, 1, "Expected to unmount the shm (and the shm only)")
})
}
func TestCleanupMountsByID(t *testing.T) {
d := &Daemon{
root: "/var/lib/docker/",
}
daemon: fix daemon.Shutdown, daemon.Cleanup not cleaning up overlay2 mounts While working on deprecation of the `aufs` and `overlay` storage-drivers, the `TestCleanupMounts` had to be updated, as it was currently using `aufs` for testing. When rewriting the test to use `overlay2` instead (using an updated `mountsFixture`), I found out that the test was failing, and it appears that only `overlay`, but not `overlay2` was taken into account. These cleanup functions were added in 05cc737f5411a0effd299429140d031c4ad8dd05, but at the time the `overlay2` storage driver was not yet implemented; https://github.com/moby/moby/tree/05cc737f5411a0effd299429140d031c4ad8dd05/daemon/graphdriver This omission was likely missed in 23e5c94cfb26eb72c097892712d3dbaa93ee9bc0, because the original implementation re-used the `overlay` storage driver, but later on it was decided to make `overlay2` a separate storage driver. As a result of the above, `daemon.cleanupMountsByID()` would ignore any `overlay2` mounts during `daemon.Shutdown()` and `daemon.Cleanup()`. This patch: - Adds a new `mountsFixtureOverlay2` with example mounts for `overlay2` - Rewrites the tests to use `gotest.tools` for more informative output on failures. - Adds the missing regex patterns to `daemon/getCleanPatterns()`. The patterns are added at the start of the list to allow for the fasted match (`overlay2` is the default for most setups, and the code is iterating over possible options). As a follow-up, we could consider adding additional fixtures for different storage drivers. Before the fix is applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 daemon_linux_test.go:135: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the shm (and the shm only) --- FAIL: TestCleanupMounts (0.01s) --- PASS: TestCleanupMounts/aufs (0.00s) --- FAIL: TestCleanupMounts/overlay2 (0.01s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 daemon_linux_test.go:171: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the root (and that only) --- FAIL: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- FAIL: TestCleanupMountsByID/overlay2 (0.00s) FAIL FAIL github.com/docker/docker/daemon 0.054s FAIL With the fix applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 --- PASS: TestCleanupMounts (0.00s) --- PASS: TestCleanupMounts/aufs (0.00s) --- PASS: TestCleanupMounts/overlay2 (0.00s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 --- PASS: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- PASS: TestCleanupMountsByID/overlay2 (0.00s) PASS ok github.com/docker/docker/daemon 0.042s Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2022-05-29 14:20:14 +00:00
t.Run("overlay2", func(t *testing.T) {
expected := "/var/lib/docker/overlay2/3a4b807fcb98c208573f368c5654a6568545a7f92404a07d0045eb5c85acaf67/merged"
var unmounted int
unmount := func(target string) error {
if target == expected {
unmounted++
}
return nil
}
err := d.cleanupMountsFromReaderByID(strings.NewReader(mountsFixtureOverlay2), "3a4b807fcb98c208573f368c5654a6568545a7f92404a07d0045eb5c85acaf67", unmount)
assert.NilError(t, err)
assert.Equal(t, unmounted, 1, "Expected to unmount the root (and that only)")
})
}
func TestNotCleanupMounts(t *testing.T) {
d := &Daemon{
repository: "",
}
var unmounted bool
unmount := func(target string) error {
unmounted = true
return nil
}
mountInfo := `234 232 0:59 / /dev/shm rw,nosuid,nodev,noexec,relatime - tmpfs shm rw,size=65536k`
daemon: fix daemon.Shutdown, daemon.Cleanup not cleaning up overlay2 mounts While working on deprecation of the `aufs` and `overlay` storage-drivers, the `TestCleanupMounts` had to be updated, as it was currently using `aufs` for testing. When rewriting the test to use `overlay2` instead (using an updated `mountsFixture`), I found out that the test was failing, and it appears that only `overlay`, but not `overlay2` was taken into account. These cleanup functions were added in 05cc737f5411a0effd299429140d031c4ad8dd05, but at the time the `overlay2` storage driver was not yet implemented; https://github.com/moby/moby/tree/05cc737f5411a0effd299429140d031c4ad8dd05/daemon/graphdriver This omission was likely missed in 23e5c94cfb26eb72c097892712d3dbaa93ee9bc0, because the original implementation re-used the `overlay` storage driver, but later on it was decided to make `overlay2` a separate storage driver. As a result of the above, `daemon.cleanupMountsByID()` would ignore any `overlay2` mounts during `daemon.Shutdown()` and `daemon.Cleanup()`. This patch: - Adds a new `mountsFixtureOverlay2` with example mounts for `overlay2` - Rewrites the tests to use `gotest.tools` for more informative output on failures. - Adds the missing regex patterns to `daemon/getCleanPatterns()`. The patterns are added at the start of the list to allow for the fasted match (`overlay2` is the default for most setups, and the code is iterating over possible options). As a follow-up, we could consider adding additional fixtures for different storage drivers. Before the fix is applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 daemon_linux_test.go:135: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the shm (and the shm only) --- FAIL: TestCleanupMounts (0.01s) --- PASS: TestCleanupMounts/aufs (0.00s) --- FAIL: TestCleanupMounts/overlay2 (0.01s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 daemon_linux_test.go:171: assertion failed: 0 (unmounted int) != 1 (int): Expected to unmount the root (and that only) --- FAIL: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- FAIL: TestCleanupMountsByID/overlay2 (0.00s) FAIL FAIL github.com/docker/docker/daemon 0.054s FAIL With the fix applied: go test -v -run TestCleanupMounts ./daemon/ === RUN TestCleanupMounts === RUN TestCleanupMounts/aufs === RUN TestCleanupMounts/overlay2 --- PASS: TestCleanupMounts (0.00s) --- PASS: TestCleanupMounts/aufs (0.00s) --- PASS: TestCleanupMounts/overlay2 (0.00s) === RUN TestCleanupMountsByID === RUN TestCleanupMountsByID/aufs === RUN TestCleanupMountsByID/overlay2 --- PASS: TestCleanupMountsByID (0.00s) --- PASS: TestCleanupMountsByID/aufs (0.00s) --- PASS: TestCleanupMountsByID/overlay2 (0.00s) PASS ok github.com/docker/docker/daemon 0.042s Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2022-05-29 14:20:14 +00:00
err := d.cleanupMountsFromReaderByID(strings.NewReader(mountInfo), "", unmount)
assert.NilError(t, err)
assert.Equal(t, unmounted, false, "Expected not to clean up /dev/shm")
}
func TestValidateContainerIsolationLinux(t *testing.T) {
d := Daemon{}
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
_, err := d.verifyContainerSettings(&configStore{}, &containertypes.HostConfig{Isolation: containertypes.IsolationHyperV}, nil, false)
assert.Check(t, is.Error(err, "invalid isolation 'hyperv' on linux"))
}
func TestShouldUnmountRoot(t *testing.T) {
for _, test := range []struct {
desc string
root string
info *mountinfo.Info
expect bool
}{
{
desc: "root is at /",
root: "/docker",
info: &mountinfo.Info{Root: "/docker", Mountpoint: "/docker"},
expect: true,
},
{
desc: "root is at in a submount from `/`",
root: "/foo/docker",
info: &mountinfo.Info{Root: "/docker", Mountpoint: "/foo/docker"},
expect: true,
},
{
desc: "root is mounted in from a parent mount namespace same root dir", // dind is an example of this
root: "/docker",
info: &mountinfo.Info{Root: "/docker/volumes/1234657/_data", Mountpoint: "/docker"},
expect: false,
},
} {
t.Run(test.desc, func(t *testing.T) {
for _, options := range []struct {
desc string
Optional string
expect bool
}{
{desc: "shared", Optional: "shared:", expect: true},
{desc: "slave", Optional: "slave:", expect: false},
{desc: "private", Optional: "private:", expect: false},
} {
t.Run(options.desc, func(t *testing.T) {
expect := options.expect
if expect {
expect = test.expect
}
if test.info != nil {
test.info.Optional = options.Optional
}
assert.Check(t, is.Equal(expect, shouldUnmountRoot(test.root, test.info)))
})
}
})
}
}
func checkMounted(t *testing.T, p string, expect bool) {
t.Helper()
mounted, err := mountinfo.Mounted(p)
assert.Check(t, err)
assert.Check(t, mounted == expect, "expected %v, actual %v", expect, mounted)
}
func TestRootMountCleanup(t *testing.T) {
if os.Getuid() != 0 {
t.Skip("root required")
}
t.Parallel()
testRoot, err := os.MkdirTemp("", t.Name())
assert.NilError(t, err)
defer os.RemoveAll(testRoot)
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
cfg := &configStore{}
err = mount.MakePrivate(testRoot)
assert.NilError(t, err)
defer mount.Unmount(testRoot)
cfg.ExecRoot = filepath.Join(testRoot, "exec")
cfg.Root = filepath.Join(testRoot, "daemon")
err = os.Mkdir(cfg.ExecRoot, 0o755)
assert.NilError(t, err)
err = os.Mkdir(cfg.Root, 0o755)
assert.NilError(t, err)
d := &Daemon{root: cfg.Root}
d.configStore.Store(cfg)
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
unmountFile := getUnmountOnShutdownPath(&cfg.Config)
t.Run("regular dir no mountpoint", func(t *testing.T) {
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
err = setupDaemonRootPropagation(&cfg.Config)
assert.NilError(t, err)
_, err = os.Stat(unmountFile)
assert.NilError(t, err)
checkMounted(t, cfg.Root, true)
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
assert.Assert(t, d.cleanupMounts(&cfg.Config))
checkMounted(t, cfg.Root, false)
_, err = os.Stat(unmountFile)
assert.Assert(t, os.IsNotExist(err))
})
t.Run("root is a private mountpoint", func(t *testing.T) {
err = mount.MakePrivate(cfg.Root)
assert.NilError(t, err)
defer mount.Unmount(cfg.Root)
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
err = setupDaemonRootPropagation(&cfg.Config)
assert.NilError(t, err)
assert.Check(t, ensureShared(cfg.Root))
_, err = os.Stat(unmountFile)
assert.Assert(t, os.IsNotExist(err))
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
assert.Assert(t, d.cleanupMounts(&cfg.Config))
checkMounted(t, cfg.Root, true)
})
// mount is pre-configured with a shared mount
t.Run("root is a shared mountpoint", func(t *testing.T) {
err = mount.MakeShared(cfg.Root)
assert.NilError(t, err)
defer mount.Unmount(cfg.Root)
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
err = setupDaemonRootPropagation(&cfg.Config)
assert.NilError(t, err)
if _, err := os.Stat(unmountFile); err == nil {
t.Fatal("unmount file should not exist")
}
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
assert.Assert(t, d.cleanupMounts(&cfg.Config))
checkMounted(t, cfg.Root, true)
assert.Assert(t, mount.Unmount(cfg.Root))
})
// does not need mount but unmount file exists from previous run
t.Run("old mount file is cleaned up on setup if not needed", func(t *testing.T) {
err = mount.MakeShared(testRoot)
assert.NilError(t, err)
defer mount.MakePrivate(testRoot)
err = os.WriteFile(unmountFile, nil, 0o644)
assert.NilError(t, err)
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
err = setupDaemonRootPropagation(&cfg.Config)
assert.NilError(t, err)
_, err = os.Stat(unmountFile)
assert.Check(t, os.IsNotExist(err), err)
checkMounted(t, cfg.Root, false)
daemon: reload runtimes w/o breaking containers The existing runtimes reload logic went to great lengths to replace the directory containing runtime wrapper scripts as atomically as possible within the limitations of the Linux filesystem ABI. Trouble is, atomically swapping the wrapper scripts directory solves the wrong problem! The runtime configuration is "locked in" when a container is started, including the path to the runC binary. If a container is started with a runtime which requires a daemon-managed wrapper script and then the daemon is reloaded with a config which no longer requires the wrapper script (i.e. some args -> no args, or the runtime is dropped from the config), that container would become unmanageable. Any attempts to stop, exec or otherwise perform lifecycle management operations on the container are likely to fail due to the wrapper script no longer existing at its original path. Atomically swapping the wrapper scripts is also incompatible with the read-copy-update paradigm for reloading configuration. A handler in the daemon could retain a reference to the pre-reload configuration for an indeterminate amount of time after the daemon configuration has been reloaded and updated. It is possible for the daemon to attempt to start a container using a deleted wrapper script if a request to run a container races a reload. Solve the problem of deleting referenced wrapper scripts by ensuring that all wrapper scripts are *immutable* for the lifetime of the daemon process. Any given runtime wrapper script must always exist with the same contents, no matter how many times the daemon config is reloaded, or what changes are made to the config. This is accomplished by using everyone's favourite design pattern: content-addressable storage. Each wrapper script file name is suffixed with the SHA-256 digest of its contents to (probabilistically) guarantee immutability without needing any concurrency control. Stale runtime wrapper scripts are only cleaned up on the next daemon restart. Split the derived runtimes configuration from the user-supplied configuration to have a place to store derived state without mutating the user-supplied configuration or exposing daemon internals in API struct types. Hold the derived state and the user-supplied configuration in a single struct value so that they can be updated as an atomic unit. Signed-off-by: Cory Snider <csnider@mirantis.com>
2022-08-31 20:12:30 +00:00
assert.Assert(t, d.cleanupMounts(&cfg.Config))
})
}
2023-01-16 22:25:13 +00:00
func TestIfaceAddrs(t *testing.T) {
CIDR := func(cidr string) *net.IPNet {
t.Helper()
nw, err := types.ParseCIDR(cidr)
assert.NilError(t, err)
return nw
}
for _, tt := range []struct {
name string
nws []*net.IPNet
}{
{
name: "Single",
nws: []*net.IPNet{CIDR("172.101.202.254/16")},
},
{
name: "Multiple",
nws: []*net.IPNet{
CIDR("172.101.202.254/16"),
CIDR("172.102.202.254/16"),
},
},
} {
t.Run(tt.name, func(t *testing.T) {
defer netnsutils.SetupTestOSContext(t)()
2023-01-16 22:25:13 +00:00
createBridge(t, "test", tt.nws...)
ipv4Nw, ipv6Nw, err := ifaceAddrs("test")
if err != nil {
t.Fatal(err)
}
assert.Check(t, is.DeepEqual(tt.nws, ipv4Nw,
cmpopts.SortSlices(func(a, b *net.IPNet) bool { return a.String() < b.String() })))
// IPv6 link-local address
assert.Check(t, is.Len(ipv6Nw, 1))
})
}
}
func createBridge(t *testing.T, name string, bips ...*net.IPNet) {
t.Helper()
link := &netlink.Bridge{
LinkAttrs: netlink.LinkAttrs{
Name: name,
},
}
if err := netlink.LinkAdd(link); err != nil {
t.Fatalf("Failed to create interface via netlink: %v", err)
}
for _, bip := range bips {
if err := netlink.AddrAdd(link, &netlink.Addr{IPNet: bip}); err != nil {
t.Fatal(err)
}
}
if err := netlink.LinkSetUp(link); err != nil {
t.Fatal(err)
}
}