df52d63854
namespaces: send config, network state and other arguments in one packet
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-30 10:38:35 +03:00
e31ef02610
Merge pull request #351 from avagin/api-rebase-2
...
Merge remote-tracking branch 'origin/master' into api-rebase
2015-01-29 19:20:09 -08:00
3c52181f61
Merge pull request #353 from LK4D4/update_dbus
...
Update github.com/godbus/dbus to v2
2015-01-29 15:46:05 -08:00
689e8ec949
Update github.com/godbus/dbus to v2
...
Signed-off-by: Alexander Morozov <lk4d4@docker.com>
2015-01-29 15:35:22 -08:00
2fac2dad91
Merge pull request #341 from shishir-a412ed/master
...
Created man page for nsinit
2015-01-29 14:00:07 -08:00
e9f8f8528a
Created man page for nsinit
...
Signed-off-by: Shishir Mahajan <shishir.mahajan@redhat.com>
2015-01-29 16:33:08 -05:00
c37b9125ec
Merge pull request #344 from hqhq/hq_fix_systemd_device
...
cgroups: always create device cgroup on systemd
2015-01-29 11:39:07 -08:00
ca633b2f29
Merge remote-tracking branch 'origin/master' into api
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-28 14:37:40 +03:00
904bae3247
Merge pull request #348 from avagin/api-nsexec
...
nsenter: remove a proxy process
2015-01-26 13:08:02 -08:00
e05f807a89
Merge pull request #349 from LK4D4/replace_wait_for_wait4
...
Use Wait4 instead of cmd.Wait
2015-01-26 13:07:54 -08:00
39fbf0a904
Use cmd.Process.Wait instead of cmd.Wait
...
Issue with cmd.Wait is that it is waiting for closing pipes and if we
have forked processes which inherited pipes from parent, then we need to
kill them to unblock cmd.Wait.
Should fix docker/docker#10303
Now idea is next:
- cmd.Process.Wait for init process dead
- Kill remaining processes in cgroup (pipes closed as side effect)
- use cmd.Wait for waiting pipes flushed to client
Signed-off-by: Alexander Morozov <lk4d4@docker.com>
2015-01-26 12:54:52 -08:00
8d8242aa8a
nsenter: add tests
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-26 23:12:50 +03:00
82367938b7
nsenter: remove a proxy process
...
Currently nsexec() creates a proxy process to enter into a pid namespace.
It isn't good, because we need to proxy an exit code and signals.
We can use CLONE_PARENT to fork a process with the right parent.
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-26 23:12:50 +03:00
11b2dab1c5
nsenter: add a macros to print errors
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-26 19:44:44 +03:00
e77b238a83
namespaces: don't send a container config twice
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-26 19:44:44 +03:00
045e9ae4a0
Merge pull request #347 from guoxiuyan/master
...
Fix a minor typo
2015-01-25 18:44:12 -08:00
7d9244eab2
Fix a minor typo
...
Signed-off-by: Guo Xiuyan <guoxiuyan@huawei.com>
2015-01-26 09:41:22 +08:00
cab4b9bce1
Merge pull request #345 from fabiokung/allow-readonly-rootfs
...
Support read-only root filesystems
2015-01-22 16:02:57 -08:00
2a452c17aa
Support read-only root filesystems
...
The only place I could find where libcontainer tries to write to the
container's root FS is when setting up the pivot dir, to be used on
pivot_root(2).
This makes the pivot base dir configurable, so a read-only FS can be
used as root FS of containers. Users can then specify a writeable
subpath to be used as pivot inside the container.
Signed-off-by: Fabio Kung <fabio@heroku.com> (github: fabiokung)
2015-01-22 13:58:41 -08:00
46573774a2
cgroups: simplify the join_memory check
...
If c.Memory=0, there is no point to set memoryswap.
Signed-off-by: Qiang Huang <h.huangqiang@huawei.com>
2015-01-22 12:19:40 +08:00
c4821b6f3e
cgroups: always create device cgroup on systemd
...
This is the same behavior as fs does.
Signed-off-by: Qiang Huang <h.huangqiang@huawei.com>
2015-01-22 09:53:30 +08:00
eb84dd1b73
add dqminh as maintainer
...
Signed-off-by: Daniel, Dao Quang Minh <dqminh89@gmail.com>
2015-01-21 20:37:37 -05:00
fae3abdadd
Merge pull request #342 from avagin/api-next
...
new-api: implement Wait, WaitProcess
2015-01-21 16:59:05 -08:00
61fef16f4a
new-api: implement Wait, WaitProcess
...
Signed-off-by: Andrew Vagin <avagin@openvz.org>
2015-01-21 18:46:01 +03:00
e79e87e426
cgroup/systemd: set config.Cgroups.Freezer
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-21 18:46:00 +03:00
6fc1dd5f25
integration: check a container state after resumning the CT
...
Otherwise CT will be left in a frozen state in a fail case
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-21 18:46:00 +03:00
5162e5a81c
integration: check container.Processes()
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-21 18:46:00 +03:00
10f0ac2921
new-api: remove nsenter.c
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-21 18:46:00 +03:00
f0c20b5601
new-api: remove DefaultCreateCommand()
...
Signed-off-by: Andrew Vagin <avagin@openvz.org>
2015-01-21 18:45:57 +03:00
3fbf185602
Merge pull request #337 from hqhq/hq_add_blkio_weight
...
add support for blkio.weight
2015-01-20 10:40:12 -08:00
5847aacb32
Merge pull request #340 from mrunalp/check_ns_flags
...
Checks namespace flags for user ns code path.
2015-01-20 10:39:23 -08:00
bde8bf2ebc
Adds namespace flag checks to userns setup.
...
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
2015-01-20 13:26:59 -05:00
107bad0ee5
Adds namespace flag checks for userns code path in init.
...
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
2015-01-20 13:26:20 -05:00
9303a8f15f
Merge pull request #324 from avagin/namespaces
...
namespace: don't change namespaces which are not belonged to the CT
2015-01-20 09:31:24 -08:00
ef73d7e235
namespace: don't change namespaces which are not belonged to the CT
...
An error is reported if a config file contains configuration for
shared namespaces.
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-20 15:05:00 +03:00
c70b6812b5
Merge pull request #339 from avagin/api-freezer
...
new-api: implement Pause() and Resume()
2015-01-19 15:17:45 -08:00
b89cd0cf5c
Merge pull request #304 from mrunalp/feature/user_namespaces
...
Adds user namespace support to libcontainer
2015-01-19 11:55:20 -08:00
b0eece8d7d
Adds support for User Namespaces.
...
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
Adds sample configuration to test user namespaces.
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
Rebases to master.
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
Fixes integration tests.
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
Move selinux labeling, apparmor profile and restrict kernel files back to init.
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
Separate the code paths for userns and default cases.
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
tty not required for setup
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
Cleanup and address review comments.
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
Remove debug logs and other cleanup.
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
Use function paramaters for SetupContainer.
Signed-off-by: Mrunal Patel <mrunalp@gmail.com> (github: mrunalp)
2015-01-19 14:50:21 -05:00
5138417f80
integration: add test to check Pause and Resume operations
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-19 22:25:52 +03:00
eb74393a3d
Merge pull request #338 from avagin/fixes
...
cgroups: set a freezer state before calling FreezerGroup.Set()
2015-01-19 09:33:08 -08:00
e451df796a
namespace: don't create needless namespaces
...
A non-zero Path field for a namespace says that a process should attach to an
existing namespace, so the process can be forked without the flag for this
namespace.
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-19 18:43:16 +03:00
02c1de6f11
cgroups: set a freezer state before calling FreezerGroup.Set()
...
My previous patch moved the setting of the freezer state after the Set()
command. It's wrong, because this command uses it, so we need to set the
freezer state before the command and rollback it in an error case.
Fixes: 13a5703
2015-01-19 17:38:26 +03:00
dcb3bca32c
namespaces: destroy cgroups only on error paths
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-19 17:38:26 +03:00
9c50d819ae
new-api: implement Pause() and Resume()
...
Signed-off-by: Andrey Vagin <avagin@openvz.org>
2015-01-19 17:38:26 +03:00
30b3306416
cgroups: set a freezer state before calling FreezerGroup.Set()
...
My previous patch moved the setting of the freezer state after the Set()
command. It's wrong, because this command uses it, so we need to set the
freezer state before the command and rollback it in an error case.
Fixes: 13a5703d85
2015-01-19 17:26:34 +03:00
54968f68bc
add support for blkio.weight
...
Signed-off-by: Qiang Huang <h.huangqiang@huawei.com>
2015-01-19 14:33:48 +08:00
73ba097bf5
Merge pull request #336 from dqminh/execin-wait
...
nsenter waits for parent signal before forking
2015-01-16 14:51:02 -08:00
5a87153824
fix TestNsenterAlivePid
...
unblock the nsenter-exec test process to let it finish succesfully
Signed-off-by: Daniel, Dao Quang Minh <dqminh89@gmail.com>
2015-01-16 05:13:19 -05:00
f5dfd9a702
nit: reindent with indent -linux
...
Signed-off-by: Daniel, Dao Quang Minh <dqminh89@gmail.com>
2015-01-16 04:58:30 -05:00
9946e299af
nsenter waits for parent signal before forking
...
this allows the parent to place the process into cgroup first so it can track
the children properly
Docker-DCO-1.1-Signed-off-by: Daniel, Dao Quang Minh <dqminh89@gmail.com> (github: dqminh)
2015-01-16 04:56:45 -05:00
Andrey Vagin
Mrunal Patel
Victor Marmol
Alexander Morozov
Rohit Jnagal
Shishir Mahajan
Michael Crosby
guoxiuyan
Fabio Kung
Qiang Huang
Daniel, Dao Quang Minh
Mrunal Patel
Alexander Morozov