Currently github.com/docker/libcontainer/cgroups/fs.GetStats does not provide the following metrics:
blkio.io_wait_time, blkio.io_service_time, blkio.io_merged, and blkio.time.
This commits add these four metrics. The descriptions of the metrics are here: https://www.kernel.org/doc/Documentation/cgroups/blkio-controller.txt.
It also modifies blkio_test and stats_util_test to ensure that the metrics are parsed.
Docker-DCO-1.1-Signed-off-by: Abin Shahab <ashahab@altiscale.com> (github: ashahab-altiscale)
For our work on adding dynamic device support to Docker we needed to be
able to call this to update the list of allowed devices. This works for
both systemd and fs based cgroups implementations.
Co-Authored-By: Chris Alfonso <calfonso@redhat.com> (github: calfonso)
Docker-DCO-1.1-Signed-off-by: Ian Main <imain@redhat.com> (github: imain)
This is needed for the libvirt exec driver.
Docker-DCO-1.1-Signed-off-by: Alexander Larsson <alexl@redhat.com> (github: alexlarsson)
Docker-DCO-1.1-Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com> (github: jpoimboe)
We calculate this on every cgroup call. It comprised of 30%+ of the CPU
usage in cAdvisor.
Docker-DCO-1.1-Signed-off-by: Victor Marmol <vmarmol@google.com> (github: vmarmol)
Sampling should be done by higher layer in configurable intervals. This
change removes the percentage stat as sampling is required for this.
Docker-DCO-1.1-Signed-off-by: Victor Marmol <vmarmol@google.com> (github: vmarmol)
This makes blkio's GetStats try to read blkio.io_serviced_recursive
first and fall back to read the stats from blkio.throttle in case it
can't. blkio.io_serviced_recursive seems to be available across all
kernels with CFQ enabled.
This closes#165
Signed-off-by: Johannes 'fish' Ziemke <github@freigeist.org>
The current paths for the different systemd cgroup subsystems that
systemd manages and that we have to manage are very inconsistent. This
patch cleans up those differences and allows consistent paths to be
used.
Signed-off-by: Michael Crosby <michael@docker.com>
Previous logic divided a uint64 by a uint64 losing precision before multiplying by clock ticks and cpus
This fix maintains precision before converting back to uint64
Docker-DCO-1.1-Signed-off-by: Nicholas Weaver <lynxbat@gmail.com> (github: lynxbat)
Docker does not require RunIn API. Hence that API has been removed.
nsinit CLI has been modified to work around the nsenter changes.
Docker-DCO-1.1-Signed-off-by: Vishnu Kannan <vishnuk@google.com> (github: vishh)
This prevents the testing package flags from leaking into the
flagsets of binaries that import libcontainer.
Docker-DCO-1.1-Signed-off-by: Peter Bourgon <peter@bourgon.org> (github: peterbourgon)
in cgroups.Cgroup is absolute. This is required to get stats of other containers while running inside a docker container.
Docker-DCO-1.1-Signed-off-by: Vishnu Kannan <vishnuk@google.com> (github: vishh)
They are really freezer state and we'll move them through another
interface.
Docker-DCO-1.1-Signed-off-by: Victor Marmol <vmarmol@google.com> (github: vmarmol)
Although it has some overlap with nsinit utility, it is useful to test
cgroup functinality directly and try out new or internal features not
exposed through nsinit.
It still needs some work as I haven't added enough systemd support yet.
I also need to expose an enter method.
Docker-DCO-1.1-Signed-off-by: Rohit Jnagal <jnagal@google.com> (github: rjnagal)
standalone cgroup test util. systemd does auto cleanup on empty, so
doesn't need an extra interface.
Docker-DCO-1.1-Signed-off-by: Rohit Jnagal <jnagal@google.com> (github: rjnagal)
This lists all currently mounted cgroups and all supported cgroup
subsystems on the machine.
Docker-DCO-1.1-Signed-off-by: Alexander Larsson <alexl@redhat.com> (github: alexlarsson)
These PR does a few things. It ensures that the freezer cgroup is
joined in the systemd driver. It also provides a public api for setting
the freezer state via the cgroups package.
Docker-DCO-1.1-Signed-off-by: Michael Crosby <michael@crosbymichael.com> (github: crosbymichael)