Ubuntu Manpages

criu

checkpoint/restore in userspace

criu <command> [options]

criu is a tool for checkpointing and restoring running applications. It does this by saving their state as a collection of files (see the dump command) and creating equivalent processes from those files (see the restore command). The restore operation can be performed at a later time, on a different system, or both.

The options are depending on the <command> criu run with.

Common options are applied to any <command>.

-v[<num>|v...]

Set logging level to <num>. The higher the level, the more output is produced. Either numeric values or multiple v can be used.

The following levels are available:

•-v1, -v only messages and errors;

•-v2, -vv also warnings (default level);

•-v3, -vvv also information messages and timestamps;

•-v4, -vvvv lots of debug.

--pidfile <file>

Write root task, service or page-server pid into a <file>.

-o, --log-file <file>

Write logging messages to <file>.

--log-pid

Write separate logging files per each pid.

-D, --images-dir <path>

Use path <path> as a base directory where to look for dump files set.

--prev-images-dir <path>

Use path <path> as a parent directory where to look for dump files set. This make sense in case of increment dumps.

-W, --work-dir <dir>

Use directory <dir> for putting logs, pidfiles and statistics. If not specified, <path> from -D option is taken.

--close <fd>

Close file with descriptor <fd> before any actions.

-L, --libdir <path>

Path to a plugins directory.

--action-script <SCRIPT>

Add an external action script. The environment variable CRTOOLS_SCRIPT_ACTION contains one of the actions:

•pre-dump run an action prior to beginning a dump

•post-dump run an action upon dump completion;

•pre-restore run an action prior to beginning a restore

•post-restore run an action upon restore completion;

•network-lock lock network in a target network namespace;

•network-unlock unlock network in a target network namespace;

•setup-namespaces run an action once root task just been created with required namespaces, note it is early stage on restore nothing were restored yet except namespaces themselves.

-V, --version

Print program version and exit.

-h, --help

Print a commands list and exit. The commands list is very short one just for overview and does not match this manual.

Launches that named pre-dump procedure, where criu does snapshot of memory changes since previous pre-dump. Also criu forms fsnotify cache which speedup restore procedure. pre-dump requires at least -t option (see dump below). Optionally page-server options may be specified.

--track-mem

Turn on memory changes tracker in the kernel. If the option is not passed the memory tracker get turned on implicitly.

Starts a checkpoint procedure.

-t, --tree <pid>

Checkpoint the whole process tree starting from <pid>.

-R, --leave-running

Leave tasks in running state after checkpoint instead of killing them. This option is pretty dangerous and should be used if and only if you understand what you are doing.

If task is about to run after been checkpointed it can modify TCP connections, delete files and do other dangerous actions. So that criu itself can not guarantee that the next restore action will not fail. Most likely if a user starts criu with this option passed at least the file system snapshot must be done with help of post-dump script.

In other words, do not use it until really needed.

-s, --leave-stopped

Leave tasks in stopped state after checkpoint instead of killing them.

-x, --ext-unix-sk [<inode>,<inode>']

Dump external unix sockets. Optionally passing <inode> (or comma separated series) it assigns inodes which allowed for one sided dump.

--freeze-cgroup

Use cgroup freezer to collect processes.

--manage-cgroups

Collect cgroups into the image thus they gonna be restored then. Without this argument criu will not save cgroups configuration associated with a task.

--cgroup-props <spec>

Specify controllers and their properties to be carried into the image file. criu predefines specifications for common controllers but since the kernel can add new controllers and modify their properties there should be a way to specify ones matched the kernel.

<spec> describes the controller and properties specification in simplified YAML form:

"c1":
 - "strategy": "merge"
 - "properties": ["a", "b"]
"c2":
 - "strategy": "replace"
 - "properties": ["c", "d"]

Where c1 and c2 are controllers names, and a, b, c, d are their properties.

Mark the format: double quotes, spaces and new lines are required. The <strategy> specify how to behave if controller specified already exist as built in one: criu either merge or replace them.

Thus for command line argument the example above will look like:

--cgroup-props "\"c1\":\n - \"strategy\": \"merge\"\n - \"properties\": [\"a\", \"b\"]\n \"c2\":\n - \"strategy\": \"replace\"\n - \"properties\": [\"c\", \"d\"]"

--cgroup-props-file <path>

Same as --cgroup-props except the specification is read from a file pointed by <path>.

--cgroup-dump-controller <name>

Dump controller with <name> only, skipping anything else automatically discovered (usually via procfs filesystem). Suitable when need criu to skip some controllers.

--cgroup-props-ignore-default

When combined with --cgroup-props makes criu to substitute predefined controller property with new one shipped. Otherwise predefined properties are merged with provided.

--tcp-established

Checkpoint established TCP connections.

--skip-in-flight

This option skips in-flight TCP connections. If TCP connections are found which are not yet completely established, criu will ignore these connections in favor of erroring out. The TCP stack on the client side is expected to handle the re-connect gracefully.

--veth-pair <IN>=<OUT>

Correspondence between outside and inside names of veth devices.

--evasive-devices

Use any path to a device file if the original one is inaccessible.

--page-server

Send pages to a page server (see page-server command).

--force-irmap

Force resolving names for inotify and fsnotify watches.

--auto-dedup

Deduplicate "old" data in pages images of previous dump. Which implies incremental dump mode (see pre-dump command).

-l, --file-locks

Dump file locks. It is necessary to make sure that all file lock users are taken into dump, so it is only safe to use this for enclosed containers where locks are not holed by someone outside of it.

-M, --ext-mount-map <KEY>:<VAL>

Setup mapping for external mounts. <KEY> is a mountpoint inside container and corresponding <VAL> is a string that will be written into the image as mountpoint's root value.

--link-remap

Allow one to link unlinked files back when possible (modifies FS till restore).

--ghost-limit size

Allow one to specify maximum allowed size of deleted file to be carried inside image files. By default up to 1M file is allowed. It is done in a sake to not carry big files inside images. size may be postfixed with K, M or G (which stands for kilo, mega and gigabytes accordingly).

-j, --shell-job

Allow one to dump shell jobs. This implies the restored task will inherit session and process group ID from the criu itself. Also this option allows one to migrate a single external tty connection, in other words this option allows one to migrate such application as "top" and friends. If passed on dump it must be specified on restore as well.

--cpu-cap [,<cap>]

Specify cap CPU capability to be written into an image file. Basically if <cap> is one of all, cpu or ins, then criu writes CPU related information into image file. If the option is omitted or set to none then image will not be written. By default criu do not write this image.

--cgroup-root [<controller>:]/<newroot>

Change the root for the controller that will be dumped. By default, CRIU simply dumps everything below where any of the tasks live. However, if a container moves all of its tasks into a cgroup directory below the container engine’s default directory for tasks, permissions will not be preserved on the upper directories with no tasks in them, which may cause problems.

Restores previously checkpointed processes.

--inherit-fd fd[<num>]:<existing>

Inherit file descriptors. This allows one to treat file descriptor <num> as being already opened via <existing> one and instead of trying to open we inherit it.

-d, --restore-detached

Detach criu itself once restore is complete.

-S, --restore-sibling

Restore root task as a sibling (make sense with --restore-detached) only.

-r, --root <path>

Change the root filesystem to <path> (when run in mount namespace).

--manage-cgroups [<mode>]

Restore cgroups configuration associated with a task from the image. Controllers are always restored in optimistic way — if already present in system criu reuses it, otherwise it will be created.

The <mode> may be one of below.

•none. Do not restore cgroup properties but require cgroup to pre-exist at the moment of restore procedure.

•props. Restore cgroup properties and require cgroup to pre-exist.

•soft. Restore cgroup properties if only cgroup has been created by criu, otherwise do not restore properties. This is the default if mode is unspecified.

•full. Always restore all cgroups and their properties.

•strict. Restore all cgroups and their properties from the scratch, requiring them to not present in the system.

--cgroup-root [<controller>:]/<newroot>

Change the root cgroup the controller will be installed into. No controller means that root is the default for all controllers not specified.

--tcp-established

Restore previously dumped established TCP connections. This implies that the network has been locked between dump and restore phases so other side of a connection simply notice a kind of lag.

--veth-pair <IN>=<OUT>

Correspondence between outside and inside names of veth devices.

-l, --file-locks

Restore file locks from the image.

-M, --ext-mount-map <KEY>:<VAL>

Setup mapping for external mounts. <KEY> is the value from the image (<VAL> from dump) and the <VAL> is the path on host that will be bind-mounted into container (to the mountpoint path from image).

--ext-mount-map auto

This is a special case. If this flag is passed, when an external mount is missing from the command line --ext-mount-map <KEY>:<VAL> syntax, criu attempts to automatically resolve this mount from its namespace.

--enable-external-sharing, --enable-external-masters

These flags enable external shared or slave mounts to be resolved automatically when --ext-mount-map auto is passed.

--auto-dedup

As soon as a page is restored it get punched out from image.

-j, --shell-job

Restore shell jobs, in other words inherit session and process group ID from the criu itself.

--cpu-cap [<cap>,<cap>]

Specify <cap> CPU capability to be present on the CPU the process is restoring. To inverse capability prefix it with ^. This option implies that --cpu-cap has been passed on dump as well, except fpu option case.

•all. Require all capabilities. This is default mode if --cpu-cap is passed without arguments. Most safe mode.

•cpu. Require the CPU to have all capabilities in image to match runtime CPU.

•fpu. Require the CPU to have compatible FPU. For example the process might be dumped with xsave capability but attempted to restore without it present on target CPU. In such case we refuse to proceed. This is default mode if --cpu-cap is not present in command line. Note this argument might be passed even if on the dump no --cpu-cap have been specified because FPU frames are always encoded into images.

•ins. Require CPU compatibility on instructions level.

•none. Ignore capabilities. Most dangerous mode. The behaviour is implementation dependent. Try to not use it until really required.

One possible need of using this option is when --cpu-cap=cpu has been passed on dump then images are migrated to a less capable processor and one need to restore this application, by default criu will refuse to proceed without relaxing capability with --cpu-cap=none parameter.

Checks whether the kernel supports the features that criu needs to successfully dump and restore a process tree.

There are three categories of kernel support as described below. criu check always checks Category 1 features unless --feature is specified which only checks the specified feature.

•Category 1. Absolutely required. These are features like /proc/<pid>/map_files, NETLINK_SOCK_DIAG socket monitoring, /proc/sys/kernel/ns_last_pid, etc.

•Category 2. Required only for specific cases. These are features like aio remap, /dev/net/tun, etc. that are required if the process being dumped or restored is using them.

•Category 3. Experimental. These are features like task-diag that are used for experimental purposes (mostly during development).

If there are no errors or warnings, criu prints "Looks good." and its exit code will be 0.

A missing Category 1 feature causes criu to print "Does not look good." and its exit code will be non-zero.

Missing Category 2 and 3 features cause criu to print "Looks good but some kernel features are missing which, depending on your process tree, may cause dump or restore failure." and its exit code will be non-zero.

Without an argument, criu check checks Category 1 features. This behavior can change with the following options:

--extra

Check kernel support for Category 2 features.

--experimental

Check kernel support for Category 3 features.

--all

Check kernel support for Category 1, 2, and 3 features.

--feature <name>

Check a specific feature. If <name> is list, a list of valid kernel feature names that can be checked will be printed.

Launches criu in page server mode.

--daemon

Runs page server as a daemon (background process).

--address <address>

Page server IP address.

--port <number>

Page server port number.

Executes a system call inside a destination task's context.

Launches criu in RPC daemon mode where criu is listening for RPC commands over socket to perform. This is convenient for the case where daemon itself is running in a privilege (superuser) mode but clients are not.

Starts pagemap data deduplication procedure, where criu scans over all pagemap files and tries to minimize the number of pagemap entries by obtaining the references from a parent pagemap image.

Fetches current CPU features and write them into an image file.

Fetches current CPU features (ie CPU the criu is running on) and test if they are compatible with ones present in image file.

To run a system call in another task's context use

    criu exec -t pid syscall-string

command. The syscall-string should look like

    syscall-name syscall-arguments ...

Each command line argument is transformed into the system call argument by the following rules:

•If one starts with &, the rest of it gets copied to the target task's address space and the respective syscall argument is the pointer to this string;

•Otherwise it is treated as a number (converted with strtol) and is directly passed into the system call.

To checkpoint a program with pid of 1234 and write all image files into directory checkpoint:

    criu dump -D checkpoint -t 1234

To restore this program detaching criu itself:

    criu restore -d -D checkpoint

To close a file descriptor number 1 in task with pid 1234:

    criu exec -t 1234 close 1

To open a file named /foo/bar for read-write in the task with pid 1234:

    criu exec -t 1234 open '&/foo/bar' 2

OpenVZ team.

Copyright (C) 2011-2015, Parallels Inc.