Runtime metrics

Docker stats

docker stats コマンドを使うと、コンテナーのランタイムメトリックスをライブ出力できます。このコマンドでは CPU、メモリー使用量、メモリー制限、ネットワーク IO メトリックスがサポートされています。

以下に docker stats コマンドの出力例を示します。

$ docker stats redis1 redis2

CONTAINER           CPU %               MEM USAGE / LIMIT     MEM %               NET I/O             BLOCK I/O
redis1              0.07%               796 KB / 64 MB        1.21%               788 B / 648 B       3.568 MB / 512 KB
redis2              0.07%               2.746 MB / 64 MB      4.29%               1.266 KB / 648 B    12.4 MB / 0 B

docker stats リファレンスページでは docker stats コマンドのより詳細について示しています。

Linux Containers rely on control groups which not only track groups of processes, but also expose metrics about CPU, memory, and block I/O usage. You can access those metrics and obtain network usage metrics as well. This is relevant for "pure" LXC containers, as well as for Docker containers.

Control groups are exposed through a pseudo-filesystem. In modern distributions, you should find this filesystem under /sys/fs/cgroup. Under that directory, you see multiple sub-directories, called devices, freezer, blkio, and so on. Each sub-directory actually corresponds to a different cgroup hierarchy.

On older systems, the control groups might be mounted on /cgroup, without distinct hierarchies. In that case, instead of seeing the sub-directories, you see a bunch of files in that directory, and possibly some directories corresponding to existing containers.

To figure out where your control groups are mounted, you can run:

$ grep cgroup /proc/mounts

Enumerate cgroups

The file layout of cgroups is significantly different between v1 and v2.

If /sys/fs/cgroup/cgroup.controllers is present on your system, you are using v2, otherwise you are using v1. Refer to the subsection that corresponds to your cgroup version.

cgroup v2 is used by default on the following distributions:

Fedora (since 31)
Debian GNU/Linux (since 11)
Ubuntu (since 21.10)

cgroup v1

You can look into /proc/cgroups to see the different control group subsystems known to the system, the hierarchy they belong to, and how many groups they contain.

You can also look at /proc/<pid>/cgroup to see which control groups a process belongs to. The control group is shown as a path relative to the root of the hierarchy mountpoint. / means the process hasn't been assigned to a group, while /lxc/pumpkin indicates that the process is a member of a container named pumpkin.

cgroup v2

On cgroup v2 hosts, the content of /proc/cgroups isn't meaningful. See /sys/fs/cgroup/cgroup.controllers to the available controllers.

Changing cgroup version

Changing cgroup version requires rebooting the entire system.

On systemd-based systems, cgroup v2 can be enabled by adding systemd.unified_cgroup_hierarchy=1 to the kernel command line. To revert the cgroup version to v1, you need to set systemd.unified_cgroup_hierarchy=0 instead.

If grubby command is available on your system (e.g. on Fedora), the command line can be modified as follows:

$ sudo grubby --update-kernel=ALL --args="systemd.unified_cgroup_hierarchy=1"

If grubby command isn't available, edit the GRUB_CMDLINE_LINUX line in /etc/default/grub and run sudo update-grub.

Running Docker on cgroup v2

Docker supports cgroup v2 since Docker 20.10. Running Docker on cgroup v2 also requires the following conditions to be satisfied:

containerd: v1.4 or later
runc: v1.0.0-rc91 or later
Kernel: v4.15 or later (v5.2 or later is recommended)

Note that the cgroup v2 mode behaves slightly different from the cgroup v1 mode:

The default cgroup driver (dockerd --exec-opt native.cgroupdriver) is systemd on v2, cgroupfs on v1.
The default cgroup namespace mode (docker run --cgroupns) is private on v2, host on v1.
The docker run flags --oom-kill-disable and --kernel-memory are discarded on v2.

Find the cgroup for a given container

For each container, one cgroup is created in each hierarchy. On older systems with older versions of the LXC userland tools, the name of the cgroup is the name of the container. With more recent versions of the LXC tools, the cgroup is lxc/<container_name>.

For Docker containers using cgroups, the container name is the full ID or long ID of the container. If a container shows up as ae836c95b4c3 in docker ps, its long ID might be something like ae836c95b4c3c9e9179e0e91015512da89fdec91612f63cebae57df9a5444c79. You can look it up with docker inspect or docker ps --no-trunc.

Putting everything together to look at the memory metrics for a Docker container, take a look at the following paths:

/sys/fs/cgroup/memory/docker/<longid>/ on cgroup v1, cgroupfs driver
/sys/fs/cgroup/memory/system.slice/docker-<longid>.scope/ on cgroup v1, systemd driver
/sys/fs/cgroup/docker/<longid>/ on cgroup v2, cgroupfs driver
/sys/fs/cgroup/system.slice/docker-<longid>.scope/ on cgroup v2, systemd driver

Metrics from cgroups: memory, CPU, block I/O

メモ
This section isn't yet updated for cgroup v2. For further information about cgroup v2, refer to the kernel documentation.

For each subsystem (memory, CPU, and block I/O), one or more pseudo-files exist and contain statistics.

Memory metrics: `memory.stat`

Memory metrics are found in the memory cgroup. The memory control group adds a little overhead, because it does very fine-grained accounting of the memory usage on your host. Therefore, many distributions chose to not enable it by default. Generally, to enable it, all you have to do is to add some kernel command-line parameters: cgroup_enable=memory swapaccount=1.

The metrics are in the pseudo-file memory.stat. Here is what it looks like:

cache 11492564992
rss 1930993664
mapped_file 306728960
pgpgin 406632648
pgpgout 403355412
swap 0
pgfault 728281223
pgmajfault 1724
inactive_anon 46608384
active_anon 1884520448
inactive_file 7003344896
active_file 4489052160
unevictable 32768
hierarchical_memory_limit 9223372036854775807
hierarchical_memsw_limit 9223372036854775807
total_cache 11492564992
total_rss 1930993664
total_mapped_file 306728960
total_pgpgin 406632648
total_pgpgout 403355412
total_swap 0
total_pgfault 728281223
total_pgmajfault 1724
total_inactive_anon 46608384
total_active_anon 1884520448
total_inactive_file 7003344896
total_active_file 4489052160
total_unevictable 32768

The first half (without the total_ prefix) contains statistics relevant to the processes within the cgroup, excluding sub-cgroups. The second half (with the total_ prefix) includes sub-cgroups as well.

Some metrics are "gauges", or values that can increase or decrease. For instance, swap is the amount of swap space used by the members of the cgroup. Some others are "counters", or values that can only go up, because they represent occurrences of a specific event. For instance, pgfault indicates the number of page faults since the creation of the cgroup.

cache: The amount of memory used by the processes of this control group that can be associated precisely with a block on a block device. When you read from and write to files on disk, this amount increases. This is the case if you use "conventional" I/O (open, read, write syscalls) as well as mapped files (with mmap). It also accounts for the memory used by tmpfs mounts, though the reasons are unclear.
rss: The amount of memory that doesn't correspond to anything on disk: stacks, heaps, and anonymous memory maps.
mapped_file: Indicates the amount of memory mapped by the processes in the control group. It doesn't give you information about how much memory is used; it rather tells you how it's used.
pgfault, pgmajfault: Indicate the number of times that a process of the cgroup triggered a "page fault" and a "major fault", respectively. A page fault happens when a process accesses a part of its virtual memory space which is nonexistent or protected. The former can happen if the process is buggy and tries to access an invalid address (it is sent a SIGSEGV signal, typically killing it with the famous Segmentation fault message). The latter can happen when the process reads from a memory zone which has been swapped out, or which corresponds to a mapped file: in that case, the kernel loads the page from disk, and let the CPU complete the memory access. It can also happen when the process writes to a copy-on-write memory zone: likewise, the kernel preempts the process, duplicate the memory page, and resume the write operation on the process's own copy of the page. "Major" faults happen when the kernel actually needs to read the data from disk. When it just duplicates an existing page, or allocate an empty page, it's a regular (or "minor") fault.
swap: The amount of swap currently used by the processes in this cgroup.
active_anon, inactive_anon: The amount of anonymous memory that has been identified has respectively active and inactive by the kernel. "Anonymous" memory is the memory that is not linked to disk pages. In other words, that's the equivalent of the rss counter described above. In fact, the very definition of the rss counter is active_anon + inactive_anon - tmpfs (where tmpfs is the amount of memory used up by tmpfs filesystems mounted by this control group). Now, what's the difference between "active" and "inactive"? Pages are initially "active"; and at regular intervals, the kernel sweeps over the memory, and tags some pages as "inactive". Whenever they're accessed again, they're immediately re-tagged "active". When the kernel is almost out of memory, and time comes to swap out to disk, the kernel swaps "inactive" pages.
active_file, inactive_file: Cache memory, with active and inactive similar to the anon memory above. The exact formula is cache = active_file + inactive_file + tmpfs. The exact rules used by the kernel to move memory pages between active and inactive sets are different from the ones used for anonymous memory, but the general principle is the same. When the kernel needs to reclaim memory, it's cheaper to reclaim a clean (=non modified) page from this pool, since it can be reclaimed immediately (while anonymous pages and dirty/modified pages need to be written to disk first).
unevictable: The amount of memory that cannot be reclaimed; generally, it accounts for memory that has been "locked" with mlock. It's often used by crypto frameworks to make sure that secret keys and other sensitive material never gets swapped out to disk.
memory_limit, memsw_limit: These aren't really metrics, but a reminder of the limits applied to this cgroup. The first one indicates the maximum amount of physical memory that can be used by the processes of this control group; the second one indicates the maximum amount of RAM+swap.

Accounting for memory in the page cache is very complex. If two processes in different control groups both read the same file (ultimately relying on the same blocks on disk), the corresponding memory charge is split between the control groups. It's nice, but it also means that when a cgroup is terminated, it could increase the memory usage of another cgroup, because they're not splitting the cost anymore for those memory pages.

CPU metrics: `cpuacct.stat`

Now that we've covered memory metrics, everything else is simple in comparison. CPU metrics are in the cpuacct controller.

For each container, a pseudo-file cpuacct.stat contains the CPU usage accumulated by the processes of the container, broken down into user and system time. The distinction is:

user time is the amount of time a process has direct control of the CPU, executing process code.
system time is the time the kernel is executing system calls on behalf of the process.

Those times are expressed in ticks of 1/100th of a second, also called "user jiffies". There are USER_HZ "jiffies" per second, and on x86 systems, USER_HZ is 100. Historically, this mapped exactly to the number of scheduler "ticks" per second, but higher frequency scheduling and tickless kernels have made the number of ticks irrelevant.

Block I/O metrics

Block I/O is accounted in the blkio controller. Different metrics are scattered across different files. While you can find in-depth details in the blkio-controller file in the kernel documentation, here is a short list of the most relevant ones:

blkio.sectors: Contains the number of 512-bytes sectors read and written by the processes member of the cgroup, device by device. Reads and writes are merged in a single counter.
blkio.io_service_bytes: Indicates the number of bytes read and written by the cgroup. It has 4 counters per device, because for each device, it differentiates between synchronous vs. asynchronous I/O, and reads vs. writes.
blkio.io_serviced: The number of I/O operations performed, regardless of their size. It also has 4 counters per device.
blkio.io_queued: Indicates the number of I/O operations currently queued for this cgroup. In other words, if the cgroup isn't doing any I/O, this is zero. The opposite is not true. In other words, if there is no I/O queued, it doesn't mean that the cgroup is idle (I/O-wise). It could be doing purely synchronous reads on an otherwise quiescent device, which can therefore handle them immediately, without queuing. Also, while it's helpful to figure out which cgroup is putting stress on the I/O subsystem, keep in mind that it's a relative quantity. Even if a process group doesn't perform more I/O, its queue size can increase just because the device load increases because of other devices.

Network metrics

Network metrics aren't exposed directly by control groups. There is a good explanation for that: network interfaces exist within the context of network namespaces. The kernel could probably accumulate metrics about packets and bytes sent and received by a group of processes, but those metrics wouldn't be very useful. You want per-interface metrics (because traffic happening on the local lo interface doesn't really count). But since processes in a single cgroup can belong to multiple network namespaces, those metrics would be harder to interpret: multiple network namespaces means multiple lo interfaces, potentially multiple eth0 interfaces, etc.; so this is why there is no easy way to gather network metrics with control groups.

Instead you can gather network metrics from other sources.

iptables

iptables (or rather, the netfilter framework for which iptables is just an interface) can do some serious accounting.

For instance, you can setup a rule to account for the outbound HTTP traffic on a web server:

$ iptables -I OUTPUT -p tcp --sport 80

There is no -j or -g flag, so the rule just counts matched packets and goes to the following rule.

Later, you can check the values of the counters, with:

$ iptables -nxvL OUTPUT

Technically, -n isn't required, but it prevents iptables from doing DNS reverse lookups, which are probably useless in this scenario.

Counters include packets and bytes. If you want to setup metrics for container traffic like this, you could execute a for loop to add two iptables rules per container IP address (one in each direction), in the FORWARD chain. This only meters traffic going through the NAT layer; you also need to add traffic going through the userland proxy.

Then, you need to check those counters on a regular basis. If you happen to use collectd, there is a nice plugin to automate iptables counters collection.

Interface-level counters

Since each container has a virtual Ethernet interface, you might want to check directly the TX and RX counters of this interface. Each container is associated to a virtual Ethernet interface in your host, with a name like vethKk8Zqi. Figuring out which interface corresponds to which container is, unfortunately, difficult.

But for now, the best way is to check the metrics from within the containers. To accomplish this, you can run an executable from the host environment within the network namespace of a container using ip-netns magic.

The ip-netns exec command allows you to execute any program (present in the host system) within any network namespace visible to the current process. This means that your host can enter the network namespace of your containers, but your containers can't access the host or other peer containers. Containers can interact with their sub-containers, though.

The exact format of the command is:

$ ip netns exec <nsname> <command...>

For example:

$ ip netns exec mycontainer netstat -i

ip netns finds the mycontainer container by using namespaces pseudo-files. Each process belongs to one network namespace, one PID namespace, one mnt namespace, etc., and those namespaces are materialized under /proc/<pid>/ns/. For example, the network namespace of PID 42 is materialized by the pseudo-file /proc/42/ns/net.

When you run ip netns exec mycontainer ..., it expects /var/run/netns/mycontainer to be one of those pseudo-files. (Symlinks are accepted.)

In other words, to execute a command within the network namespace of a container, we need to:

Find out the PID of any process within the container that we want to investigate;
Create a symlink from /var/run/netns/<somename> to /proc/<thepid>/ns/net
Execute ip netns exec <somename> ....

Review Enumerate Cgroups for how to find the cgroup of an in-container process whose network usage you want to measure. From there, you can examine the pseudo-file named tasks, which contains all the PIDs in the cgroup (and thus, in the container). Pick any one of the PIDs.

Putting everything together, if the "short ID" of a container is held in the environment variable $CID, then you can do this:

$ TASKS=/sys/fs/cgroup/devices/docker/$CID*/tasks
$ PID=$(head -n 1 $TASKS)
$ mkdir -p /var/run/netns
$ ln -sf /proc/$PID/ns/net /var/run/netns/$CID
$ ip netns exec $CID netstat -i

Tips for high-performance metric collection

Running a new process each time you want to update metrics is (relatively) expensive. If you want to collect metrics at high resolutions, and/or over a large number of containers (think 1000 containers on a single host), you don't want to fork a new process each time.

Here is how to collect metrics from a single process. You need to write your metric collector in C (or any language that lets you do low-level system calls). You need to use a special system call, setns(), which lets the current process enter any arbitrary namespace. It requires, however, an open file descriptor to the namespace pseudo-file (remember: that's the pseudo-file in /proc/<pid>/ns/net).

However, there is a catch: you must not keep this file descriptor open. If you do, when the last process of the control group exits, the namespace isn't destroyed, and its network resources (like the virtual interface of the container) stays around forever (or until you close that file descriptor).

The right approach would be to keep track of the first PID of each container, and re-open the namespace pseudo-file each time.

Collect metrics when a container exits

Sometimes, you don't care about real time metric collection, but when a container exits, you want to know how much CPU, memory, etc. it has used.

Docker makes this difficult because it relies on lxc-start, which carefully cleans up after itself. It is usually easier to collect metrics at regular intervals, and this is the way the collectd LXC plugin works.

But, if you'd still like to gather the stats when a container stops, here is how:

For each container, start a collection process, and move it to the control groups that you want to monitor by writing its PID to the tasks file of the cgroup. The collection process should periodically re-read the tasks file to check if it's the last process of the control group. (If you also want to collect network statistics as explained in the previous section, you should also move the process to the appropriate network namespace.)

When the container exits, lxc-start attempts to delete the control groups. It fails, since the control group is still in use; but that's fine. Your process should now detect that it is the only one remaining in the group. Now is the right time to collect all the metrics you need!

Finally, your process should move itself back to the root control group, and remove the container control group. To remove a control group, just rmdir its directory. It's counter-intuitive to rmdir a directory as it still contains files; but remember that this is a pseudo-filesystem, so usual rules don't apply. After the cleanup is done, the collection process can exit safely.