Patch to the lxc-fedora template to setup gettys on the ttys that are
enabled in the configuration. The area of the code already had some
modifications to that service that didn't seem to do anything and would
get wiped out by an update. I commented that out but subsumed the
change it was attempting into my command in case it does something on
another rev somewhere.
This is very similar to the logic in the OpenSuse template but doesn't
seem to appear in other templates, such as arch, which have to deal with
systemd. This isn't unique to Fedora. The templates for Fedora,
ArchLinux, and OpenSuse are the only three that seem to have any
reference to systemd at all.
Attached below the jump.
Regards,
Mike
--
Michael H. Warfield (AI4NB) | (770) 985-6132 | mhw@WittsEnd.com
/\/\|=mhw=|\/\/ | (678) 463-0932 | http://www.wittsend.com/mhw/
NIC whois: MHW9 | An optimist believes we live in the best of all
PGP Key: 0x674627FF | possible worlds. A pessimist is sure of it!
--
Signed-off-by: Michael H. Warfield <mhw@WittsEnd.com> Signed-off-by: Serge Hallyn <serge.hallyn@ubuntu.com>
Dwight Engen [Fri, 17 May 2013 22:28:12 +0000 (18:28 -0400)]
return lxc generated name for veth pair
Doing a get_config_item for lxc.network.0.veth.pair only returns the
pair name if explicitly given, but it can be useful to know the name
even if it is the one that lxc autogenerated.
Serge Hallyn [Tue, 14 May 2013 21:10:37 +0000 (16:10 -0500)]
lxc: add clone hook.
Add a clone hook called from api_clone. Pass arguments to it from
lxc_clone.c.
The clone update hook is called while the container's bdev is mounted.
Information about the container is passed in through environment
variables LXC_ROOTFS_PATH, LXC_NAME, The LXC_ROOTFS_MOUNT, and
LXC_CONFIG_FILE.
So from the hook, updates to the container should be made under
$LXC_ROOTFS_MOUNT/ .
The hook also receives command line arguments as follows:
First argument is container name, second is always 'lxc', third
is the hook name (always clone), then come the arguments which
were passed to lxc-clone. I.e. when I did:
sudo lxc-clone demo2 demo3 -- hey there dude
the arguments passed in were "demo3 lxc clone hey there dude"
I personally would like to drop the first two arguments. The
name is available as $LXC_NAME, and the section argument ('lxc')
is meaningless. However, doing so risks invalidating existing
hooks.
Soon analogous create and destroy hooks will be added as well.
Serge Hallyn [Wed, 15 May 2013 20:21:24 +0000 (15:21 -0500)]
cgroup: prevent DOS when a hierachy is mounted multiple times
When starting a container, we walk through all cgroup mounts looking
for a unique directory name we can use for this container. If the
name we are trying is in use, we try another name. If it is not in
use in the first mount we check, we need to check other hierarchies
as it may exist there. But we weren't checking whether we have already
checked a subsystem - so that if freezer was mounted twice, we would
create it in the first mount, see it exists in the second, so start
over trying in the second mount.
To fix this, keep track of which subsystems we have already checked,
and do not re-check.
Note we still need to add, at the next: label, the removal of the
directories we've already created. I'm keeping that for later as
it's far lower priority than this fix, and I don't want to risk
introducing a regression for that.
Dwight Engen [Wed, 15 May 2013 16:27:34 +0000 (12:27 -0400)]
set non device cgroup items before the cgroup is entered
This allows some special cgroup items such as memory.kmem.limit_in_bytes
to be successfully set, since they must be set before any task is put
into the cgroup.
The devices cgroup is setup later giving the container a chance to mount
file systems before the device it might want to mount from becomes
unavailable.
lxc-fedora-template: autodev, hostname, ARM archs, Raspberry Pi fixes
This took a lot longer for me to get around to it... Sorry.
Patch to the lxc-fedora template.
I didn't get any further comments from my earlier proposal, weeks ago,
and did get one addition based on comments about properly setting the
hostname in /etc/hostname, which I've added. I could have broken them
into separate patches but most are pretty small and minor.
Changes:
* Map armv6l and armv7l architectures to "arm" for yum and repos to
function properly.
* Detect Fedora Remix distros with no "/etc/fedora-release" file
(Raspberry Pi) and find proper release versions when "remix" part of the
file context.
* Change default Fedora container on non-Fedora hosts to Fedora 17.
* Added code for autodev for Fedora systemd containers.
* Added code to set /etc/hostname for Fedora > 14 (systemd).
* Fix a few typos.
Regards,
Mike
--
Michael H. Warfield (AI4NB) | (770) 985-6132 | mhw@WittsEnd.com
/\/\|=mhw=|\/\/ | (678) 463-0932 | http://www.wittsend.com/mhw/
NIC whois: MHW9 | An optimist believes we live in the best of all
PGP Key: 0x674627FF | possible worlds. A pessimist is sure of it!
--
Signed-off-by: Michael H. Warfield <mhw@WittsEnd.com> Signed-off-by: Serge Hallyn <serge.hallyn@ubuntu.com>
lxc-busybox: check when bind-mounting host libdirs
The patch removes the behavior of automatically mounting /lib
and /usr/lib, since this is duplicated a few lines below. It will
also remove the risk of failing when one of these entries are not
present on the host - e.g. on a 64bit machine.
Serge Hallyn [Fri, 10 May 2013 05:52:22 +0000 (00:52 -0500)]
add lxc-cirros
Add a template to create a cirros container. One great thing about
cirros is that the image you download is 3.5M.
Thanks smoser!
Note by default /etc/inittab doesn't have a /dev/console entry, so you
don't get a login on the lxc-start console. Adding
console::respawn:/sbin/getty 115200 console
makes that work, but ctrl-c still gets forwarded to init which then
reboots. So I didn't bother adding console as part of the template
(yet). Instead I simply lxc-start -d, then lxc-console.
Signed-off-by: Scott Moser <scott.moser@canonical.com> Signed-off-by: Serge Hallyn <serge.hallyn@ubuntu.com>
Dwight Engen [Mon, 13 May 2013 16:03:14 +0000 (12:03 -0400)]
serialize multiple threads doing lxcapi_start()
The problem is that the fd table is shared between threads and if a thread
forks() while another thread has an open fd to the monitor, the duped fd
in the fork()ed child will not get closed, thus causing monitord to stay
around since it thinks it still has a client. This only happened when
calling lxcapi_start() in the daemonized case since that is the only time
we try to get the status from the monitor.
Serge Hallyn [Wed, 8 May 2013 00:28:32 +0000 (19:28 -0500)]
lxc-ps: handle cgroup collisions
A few months ago cgroup handling in lxc was updated so that if
/sys/fs/cgroup/$cgroup/lxc/$container already exists (most often
due to another container by the same name under a different lxcpath),
then /sys/fs/cgroup/$cgroup/lxc/${container}-N would be used.
lxc-ps was never updated to handle this. Fix that.
(Note, the ns cgroup is being special cased there, but I don't
really believe ns cgroup works any more.)
It would be preferable to rewrite lxc-ps in python or in C, but
this at least makes the basic lxc-ps work in the case of multiple
containers with the same name.
Changelog:
fix missing fi.
replace 'z1' with '$container' as pointed out by Christian
Serge Hallyn [Tue, 7 May 2013 20:33:42 +0000 (15:33 -0500)]
conf.c: remove a break
commit ab81cef05338e7a553aacca141287034d6daf167 meant to remove the
added break, but apparently i had not done 'git add' before commit
--amend. Remove the added break.
Dwight Engen [Fri, 3 May 2013 17:41:40 +0000 (13:41 -0400)]
coverity: ftell returns a signed value
The check for flen < 0 could never have been true since flen was declared
to be size_t (unsigned). Declare flen to be long since that is what ftell
returns.
Dwight Engen [Fri, 3 May 2013 16:04:01 +0000 (12:04 -0400)]
coverity: fix leak in error case
Since lxc_execute() is available through the library and is exposed via
the API we cannot be sure the caller will immediately exit, so we should
take care to free the allocated memory.
Weng Meiling [Fri, 3 May 2013 03:02:48 +0000 (11:02 +0800)]
lxc_start: free the conf if starting the container fails
When running lxc-start command with valgrind, it reports a memory leak error.
When lxc-start command fails, the conf which is from malloc has not been released.
This patch fix the problem.
> Quoting Dwight Engen (dwight.engen@oracle.com):
> > So I did this, only to realize that lxc-init is passing "none" for
> > the file anyway, so it currently doesn't intend to log. This makes
> > me think that passing NULL for lxcpath is the right thing to do in
> > this patch. If you want me to make it so lxc-init can log, I can do
> > that but I think it should be in a different change :)
>
> That actually would be very useful, but as you say that's a different
> feature - thanks.
It's a tiny program (exported through the api) wrapping the util.c
helpers for reading /etc/lxc/lxc.conf variables, and replaces
the kludgy shell duplication in lxc.functions.in
Changelog: Apr 30: address feedback from Dwight
(exit error on failure, and use 'lxcpath' as name, not
'default_path').
for lvm and zfs, as we don't yet support passing options, only default
VG of 'lxc' and default zfsroot of 'tank' are supported when converting
another backing store type.
refuse deletion of container which has lvm or zfs snapshots.
Note that since a zfs clone must be made from a zfs snapshot,
which is made from the original zfs fs, even after we
lxc-destroy the snapshotted container we still must manually
remove the snapshot. This can be handled automatically, by
looking for snapshots where c1 is the original, c2 is the clone,
tank/c2 no longer exists, but tank/c1@c2 does. We can then
remove tank/c1@c2 and feel free to remove tank/c1. This patch
does NOT do that yet.
Make sure not to return when we're a forked child.
implement backend drivers and container clone API (v3)
1. commonize waitpid users to use a single helper. We frequently want
to run something in a clean namespace, or fork off a script. This
lets us keep the function doing fork:(1)exec(2)waitpid simpler.
2. start a blockdev backend implementation. This will be used for
mounting, copying, and snapshotting container filesystems.
3. implement btrfs, lvm, directory, and overlayfs backends.
4. For overlayfs, support a new lxc.rootfs format of
'bdevtype:<extra>'. This means you can now use overlayfs-based
containers without using lxc-start-ephemeral, by using
lxc.rootfs = overlayfs:/readonly-dir:writeable-dir
5. add a set of simple clone testcases
6. Write a new lxc_clone.c based on api clone.
Still to do (there's more, but off top of my head):
1. support zfs, aufs
2. have clone handle other mount entries (right now it only clones
the rootfs)
3. python, lua, and go bindings (not me :)
4. lxc-destroy: if lvm backing store, check for snapshots of it.
(what about directories which have overlayfs clones?)
Changes since v2:
Initialize random generator when picking new macaddr (reported
by caglar@10ur.org)
Fix wrong use of bitmask flags
On copy-clone of btrfs, create a subvolume
lxc_clone.c: respect the command line usage of the old script
lxc-clone(1): update documentation
Refuse to try changing backing stores expect to overlayfs, as
it is not implemented (yet) anyway.
> Quoting Dwight Engen (dwight.engen@oracle.com):
> > On Fri, 26 Apr 2013 09:37:49 -0500
> > Serge Hallyn <serge.hallyn@ubuntu.com> wrote:
> >
> > > Quoting Dwight Engen (dwight.engen@oracle.com):
> > > > Using lxc configured with --enable-configpath-log, and
> > > > specifying a path to the lxc commands with -P, the log file
> > > > path is generated with a basename of LOGPATH instead of the
> > > > lxcpath. This means for example if you do
> > > >
> > > > lxc-start -P /tmp/containers -n test01 -l INFO
> > > >
> > > > your log file will be
> > > >
> > > > /var/lib/lxc/test01/test01.log
> > > >
> > > > I was expecting the log to be /tmp/containers/test01/test01.log.
> > > > This is particularly confusing if you also have test01 on the
> > > > regular lxcpath. The patch below changes the log file path to be
> > > > based on lxcpath rather than LOGPATH when lxc is configured with
> > > > --enable-configpath-log.
> > > >
> > > > I think that even in the normal non --enable-configpath-log case
> > > > we should consider using lxcpath as the base and not having
> > > > LOGPATH at all, as attempting to create the log files
> > > > in /var/log is not going to work for regular users on their own
> > > > lxcpath. If we want that, I'll update the patch to do that as
> > > > well.
> > >
> > >
> > > Perhaps we should do:
> > >
> > > 1. If lxcpath == default_lxc_path(), then first choice is
> > > LOGPATH, second is lxcpath/container.log
> > > 2. when opening, if first choice fails, use second choice
> > > if there is any.
> > >
> > > That way 'system' containers will go to /var/log/lxc, as I think
> > > they should. Custom-lxcpath containers should never go
> > > to /var/log/lxc, since their names could be dups of containers in
> > > default_lxc_path(). And if the system is a weird one where
> > > default_lxc_path is set up so that an unprivileged user can use
> > > it, then we should log into $lxcpath.
> >
> > That sounds good to me. So these rules would apply in both the
> > regular and --enable-configpath-log cases.
I updated the patch to try to open the log file according to the
choices given above. Along the way I cleaned up log.c a bit, making
some things static, grouping external interfaces together, etc...
Hopefully that doesn't add too much noise.
> > > (Note this patch will trivially conflict with my new lxc_clone.c
> > > causing it to fail to build - unfortunate result of timing)
> >
> > Yeah unfortunately this touches every lxc_log_init() caller. I can
> > work on the above logic and re-submit after your new lxc_clone
> > stuff goes in.
>
> No no, I'll just need to remember to update mine. Don't hold up on
> mine, this is just the nature of such collaboration :)
>
> > Did you have any thoughts on the XXX what to pass in for lxcpath in
> > lxc_init? Right now it just falls back to LOGPATH.
>
> No - that's a weird one, since lxc_init runs in the container. If
> there were only system containers I'd say always use LOGPATH.
> However there are people (apparently :) who use container sharing the
> host's rootfs...
>
> lxc-execute does know the lxcpath. Perhaps we can simply have
> src/lxc/execute.c:execute_start() look at handler->conf to see if a
> rootfs is set. If rootfs is NOT set, then pass lxcpath along to
> lxc-init. Then lxc-init can mostly do the same as the others? (It
> doesn't use src/lxc/arguments.c, so you'd have to add lxcpath to
> options[] in lxc-init.c)
So I did this, only to realize that lxc-init is passing "none" for the
file anyway, so it currently doesn't intend to log. This makes me
think that passing NULL for lxcpath is the right thing to do in
this patch. If you want me to make it so lxc-init can log, I can do
that but I think it should be in a different change :)
Commit 69fe23ff added checking for the older docbook2man back into
configure, but this breaks building the docs on at least Oracle Linux and
Fedora when docbook2X is not installed as docbook2man will be found but the
docs don't actually build with that tool.
This change makes it so the docs can be built with either the older
docbook2man or the newer 2X tools by using configure to set the dtd
string to an appropriate value depending on use of docbook2man or
db2x_docbook2man.
Also fixed a small error in lxc-destroy.sgml.in that was noticed
by the old tools.
I'll write the bdev.c handler and hook up lxc-clone next.
This also fixses a bug in the sed expression to extract the rootfs from
container config, which prepended an extra '/' to the rootdev. (That
caused the zfs list entry not to match at destroy)
If lxc_wait is called before the container has started the socket will not
yet have been created and lxc_wait's connect to it will fail. Starting the
daemon will create the socket for lxc_wait to connect to.
Signed-off-by: Dwight Engen <dwight.engen@oracle.com> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
Christian Seiler [Thu, 25 Apr 2013 11:00:19 +0000 (13:00 +0200)]
lxc_attach: Use clone() instead of second fork()
Because of an assertion in glibc's fork() wrapper that parent pid and
pid of child should never be the same, one should avoid fork() after
attaching to a PID namespace, since the pid inside the namespace may
coincide with the pid of the parent outside the namespace, thus hitting
the aforementioned assertion.
This patch just changes the code in the most simple manner to use
clone() instead of fork(). Since clone() requires a function to be
called instead of returning 0, we move the code of the child into a
function child_main.
Signed-off-by: Christian Seiler <christian@iwakd.de> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
Peter Simons [Thu, 25 Apr 2013 10:20:30 +0000 (12:20 +0200)]
configure: support for the "docbook2man" utility to build the documentation
This adds docbook2man as an alternative name for the docbook compiler.
As that name was used on Debian based systems for an older version of the tool,
this change also adds a check so that docbook2man is never used on Debian based
systems.
Reported-by: Peter Simons <simons@cryp.to> Reported-by: Christian Bühler christian@cbuehler.de Signed-off-by: Stéphane Graber <stgraber@ubuntu.com> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
This fixes a long standing issue that there could only be a single
lxc-monitor per container.
With this change, a new lxc-monitord daemon is spawned the first time
lxc-monitor is called against the container and will accept connections
from any subsequent lxc-monitor.
Signed-off-by: Dwight Engen <dwight.engen@oracle.com> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
Trying to start multiple containers concurrently may cause
lxc_monitor_read_timeout to fail as select call could be
interrupted by a signal, handle it.
Trying to stop multiple containers concurrently ends up with "cgroup is not mounted" errors as multiple threads corrupts the shared variables.
Fix that stack corruption and start to use getmntent_r to support stopping multiple containers concurrently.
Signed-off-by: S.Çağlar Onur <caglar@10ur.org> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
Introduce a new HTTP_PROXY variable in /etc/default/lxc. If unset or
set to none, then behavior continues as before. If set to 'apt', then
any http::proxy set in apt.conf will be used as http_proxy for
debootstrap, and specified in the container's
/etc/apt/apt.conf.d/70proxy. If set to something else, then the
value of HTTP_PROXY will be used as http_proxy for debootstrap and
specified in the container's 70proxy.
Changelog: (apr 23) merge the two apt proxy detection functions.
When using -P (lxcpath), the parameter path needs to be forwarded
to the various commands being run but not used by the nested lxc-ls
as it's relatively unlikely that both the host and the nested containers
use a custom path.
This isn't ideal but short of having a way to provide the container path
for every single of the nesting (with potential unlimited depth), it's
the best we can do.
Signed-off-by: Stéphane Graber <stgraber@ubuntu.com> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
- Drop disabled entries from allowed devices list
- Improve generated config layout a bit
- Drop redundant uname call
- Re-generate the SSH host keys on container creation
Signed-off-by: Stéphane Graber <stgraber@ubuntu.com> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
This fixes a few issues uncovered by the recent C module fix.
In lxc-start-ephemeral, the hwaddr code wasn't actually working.
Replace by code that properly iterates through the network interfaces
and sets a new MAC address for each entry.
In the python overlay, catch the newly emitted KeyError when in
set_config_item (or setting any previously unset variable would fail).
Signed-off-by: Stéphane Graber <stgraber@ubuntu.com> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
Fixes a lot of issues found by a code review done by Barry Warsaw.
Those include:
- Wrong signature for getters
- Various memory leaks
- Various optimizations
- More consistent return values
- Proper exception handling
Signed-off-by: Stéphane Graber <stgraber@ubuntu.com> Reported-by: Barry Warsaw <barry@ubuntu.com> Acked-by: Barry Warsaw <barry@ubuntu.com> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
Reimplement mkdir_p() such that it:
...handles relativ paths correctly. (currently it crashes)
...does not rely on dirname().
...is not recursive.
...is shorter. ;-)
Signed-off-by: Richard Weinberger <richard@nod.at> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
This commit was preventing startup of containers using lxc hooks and
shutdown of all other containers, requiring the use of a good old
kill -9 to get rid of lxc-start after a container shutdown.
Reimplement mkdir_p() such that it:
...handles relativ paths correctly. (currently it crashes)
...does not rely on dirname().
...is not recursive.
...is shorter. ;-)
Signed-off-by: Richard Weinberger <richard@nod.at> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
If the process in the new namespace dies very early
we have currently no chance to detect this.
The parent process will just die due to SIGPIPE
if it write to the fd used for synchronisation and
nobody will notice the real cause of the problem.
Install a SIGCHLD handler to detect the death.
Later when the child does execve() to the init within
the new namespace the handler will be disabled automatically.
Signed-off-by: Richard Weinberger <richard@nod.at> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
quiet gcc 4.4.7 warning about saveptr use before initialization
The recent change to use strtok_r causes a build warning with this older
gcc version, so initialize saveptr to NULL to quiet the compiler and
unbreak the build. There was no warning with gcc 4.7.2 that I
originally tested with.
As Richard reported, dirname('//') returns //. But mkdir_p only stops
when called with '/', resulting in infinite recursion when given a
pathname '//foo/bar'.
Reported-by: richard -rw- weinberger <richard.weinberger@gmail.com> Signed-off-by: Serge Hallyn <serge.hallyn@ubuntu.com>
lxc-template: enable chroot + chpasswd functionality for Busybox hosts
This patch supports the scenario where a user wants to install a
busybox container on a busybox host.
When running the template, in order to change the root password,
the template needs to do the chroot. On busybox-powered hosts, chroot
is not part of the coreutils package - it's part of busybox. And the
busybox implementation or chroot only works if it has /lib in the new
root populated with the right binaries (or at least that's the
solution I found to make it work).
The temporarily bind-mounts /lib in the NEWROOT, chroots there,
changes the password, goes back and unmounts. This set of operations
is contained in a new MOUNT namespace, using the lxc-unshare call.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@freescale.com> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
pclose returns the exit status from wait, we need to check that to see if
the script itself failed or not. Tested a script that returned 0, 1, and
also one that did a sleep and then was killed by a signal.
Signed-off-by: Dwight Engen <dwight.engen@oracle.com> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
Also check that we wrote the amount we expected to. The write on the pty
is blocking but we could still get a short write on EINTR, so we should
SYSERROR it.
Signed-off-by: Dwight Engen <dwight.engen@oracle.com> Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
projects / thirdparty / lxc.git / log
