Not using native diff for overlay2

[cespo]

Description of problem

Starting a docker:dind container with the kata-runtime, the 'Extracting' phase of the nested docker pull commands is very slow.

In the logs I can find an error saying:

level=warning msg="Not using native diff for overlay2, this may cause degraded performance for building images: failed to set opaque flag on middle layer: operation not supported" storage-driver=overlay2

Expected result

Docker commands in kata-runtime to run at the same speed of non kata-runtime

Actual result

Docker commands in kata-runtime are very slow

From within the dind container:

# docker info
Server Version: 18.06.3-ce
Storage Driver: overlay2
 Backing Filesystem: extfs
 Supports d_type: true
 Native Overlay Diff: false

---

Pod Spec

apiVersion: v1
kind: Pod
metadata:
  name: dind
  annotations:
    io.kubernetes.cri.untrusted-workload: "true"
spec:
  containers:
  - name: dind
    securityContext:
      privileged: true
    image: docker:18.06-dind
    args: ["--storage-driver=overlay2"]
    volumeMounts:
      - mountPath: /var/run/
        name: dockersock
  volumes:
    - name: dockersock
      emptyDir: {}

---

Kernel config file

# CONFIG_OVERLAY_FS_REDIRECT_DIR is not set
# CONFIG_OVERLAY_FS_REDIRECT_ALWAYS_FOLLOW is not set
# CONFIG_OVERLAY_FS_INDEX is not set
# CONFIG_OVERLAY_FS_XINO_AUTO is not set
# CONFIG_OVERLAY_FS_METACOPY is not set

I tried different configurations with CONFIG_OVERLAY_FS_REDIRECT_DIR and CONFIG_OVERLAY_FS_METACOPY on/off as well as ALL the overlay_fs modules on. Same error

---

Containerd

I have built containerd from master to include mxpv:devmapper branch

TAG='v1.2.0-349-g9ab4c8c'

go get github.com/containerd/containerd
pushd $GOPATH/src/github.com/containerd/containerd

git checkout tags/${TAG} -b custom/containerd-${TAG}
 
git remote add mxpv https://github.com/mxpv/containerd
git fetch mxpv devmapper
git rebase custom/containerd-${TAG} mxpv/devmapper

---

Meta details

Running kata-collect-data.sh version 1.6.0 (commit 3c3d0fbf5345f6ca01bb32d81a2d3c73bd2edff0) at 2019-03-27.00:37:59.206079458+0000.

---

Runtime is /opt/kata/bin/kata-runtime.

kata-env

Output of "/opt/kata/bin/kata-runtime kata-env":

[Meta]
  Version = "1.0.20"

[Runtime]
  Debug = false
  Trace = false
  DisableGuestSeccomp = true
  DisableNewNetNs = false
  Path = "/opt/kata/bin/kata-runtime"
  [Runtime.Version]
    Semver = "1.6.0"
    Commit = "3c3d0fbf5345f6ca01bb32d81a2d3c73bd2edff0"
    OCI = "1.0.1-dev"
  [Runtime.Config]
    Path = "/etc/kata-containers/configuration.toml"

[Hypervisor]
  MachineType = "pc"
  Version = "QEMU emulator version 2.11.2(kata-static)\nCopyright (c) 2003-2017 Fabrice Bellard and the QEMU Project developers"
  Path = "/opt/kata/bin/qemu-system-x86_64"
  BlockDeviceDriver = "virtio-scsi"
  EntropySource = "/dev/urandom"
  Msize9p = 8192
  MemorySlots = 10
  Debug = false
  UseVSock = false

[Image]
  Path = "/opt/kata/share/kata-containers/kata-containers-ubuntu.img"

[Kernel]
  Path = "/opt/kata/share/kata-containers/kitt-2.0.0_kata-vmlinuz-4.19.24.container"
  Parameters = "init=/usr/lib/systemd/systemd systemd.unit=kata-containers.target systemd.mask=systemd-networkd.service systemd.mask=systemd-networkd.socket systemd.mask=systemd-journald.service systemd.mask=systemd-journald.socket systemd.mask=systemd-journal-flush.service systemd.mask=systemd-udevd.service systemd.mask=systemd-udevd.socket systemd.mask=systemd-udev-trigger.service systemd.mask=systemd-timesyncd.service systemd.mask=systemd-update-utmp.service systemd.mask=systemd-tmpfiles-setup.service systemd.mask=systemd-tmpfiles-cleanup.service systemd.mask=systemd-tmpfiles-cleanup.timer systemd.mask=tmp.mount"

[Initrd]
  Path = ""

[Proxy]
  Type = "kataProxy"
  Version = "kata-proxy version 1.6.0-91fd43745d4b3e04f1b74714172e16f74c96d57b"
  Path = "/opt/kata/libexec/kata-containers/kata-proxy"
  Debug = false

[Shim]
  Type = "kataShim"
  Version = "kata-shim version 1.6.0-a2e583b73441f60f9191390fb3ff44194407cfef"
  Path = "/opt/kata/libexec/kata-containers/kata-shim"
  Debug = false

[Agent]
  Type = "kata"

[Host]
  Kernel = "4.19.23-coreos-r1"
  Architecture = "amd64"
  VMContainerCapable = true
  SupportVSocks = false
  [Host.Distro]
    Name = "Container Linux by CoreOS"
    Version = "2023.4.0"
  [Host.CPU]
    Vendor = "GenuineIntel"
    Model = "Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz"

[Netmon]
  Version = "kata-netmon version 1.6.0"
  Path = "/opt/kata/libexec/kata-containers/kata-netmon"
  Debug = false
  Enable = false

---

Runtime config files

Runtime default config files

/etc/kata-containers/configuration.toml
/opt/kata/share/defaults/kata-containers/configuration.toml

Runtime config file contents

Output of "cat "/etc/kata-containers/configuration.toml"":

# Copyright (c) 2017-2019 Intel Corporation
#
# SPDX-License-Identifier: Apache-2.0
#

# XXX: WARNING: this file is auto-generated.
# XXX:
# XXX: Source file: "cli/config/configuration-qemu.toml.in"
# XXX: Project:
# XXX:   Name: Kata Containers
# XXX:   Type: kata

[hypervisor.qemu]
path = "/opt/kata/bin/qemu-system-x86_64"
kernel = "/opt/kata/share/kata-containers/vmlinuz.container"
image = "/opt/kata/share/kata-containers/kata-containers.img"
machine_type = "pc"

# Optional space-separated list of options to pass to the guest kernel.
# For example, use `kernel_params = "vsyscall=emulate"` if you are having
# trouble running pre-2.15 glibc.
#
# WARNING: - any parameter specified here will take priority over the default
# parameter value of the same name used to start the virtual machine.
# Do not set values here unless you understand the impact of doing so as you
# may stop the virtual machine from booting.
# To see the list of default parameters, enable hypervisor debug, create a
# container and look for 'default-kernel-parameters' log entries.
kernel_params = ""

# Path to the firmware.
# If you want that qemu uses the default firmware leave this option empty
firmware = ""

# Machine accelerators
# comma-separated list of machine accelerators to pass to the hypervisor.
# For example, `machine_accelerators = "nosmm,nosmbus,nosata,nopit,static-prt,nofw"`
machine_accelerators=""

# Default number of vCPUs per SB/VM:
# unspecified or 0                --> will be set to 1
# < 0                             --> will be set to the actual number of physical cores
# > 0 <= number of physical cores --> will be set to the specified number
# > number of physical cores      --> will be set to the actual number of physical cores
default_vcpus = 1

# Default maximum number of vCPUs per SB/VM:
# unspecified or == 0             --> will be set to the actual number of physical cores or to the maximum number
#                                     of vCPUs supported by KVM if that number is exceeded
# > 0 <= number of physical cores --> will be set to the specified number
# > number of physical cores      --> will be set to the actual number of physical cores or to the maximum number
#                                     of vCPUs supported by KVM if that number is exceeded
# WARNING: Depending of the architecture, the maximum number of vCPUs supported by KVM is used when
# the actual number of physical cores is greater than it.
# WARNING: Be aware that this value impacts the virtual machine's memory footprint and CPU
# the hotplug functionality. For example, `default_maxvcpus = 240` specifies that until 240 vCPUs
# can be added to a SB/VM, but the memory footprint will be big. Another example, with
# `default_maxvcpus = 8` the memory footprint will be small, but 8 will be the maximum number of
# vCPUs supported by the SB/VM. In general, we recommend that you do not edit this variable,
# unless you know what are you doing.
default_maxvcpus = 0

# Bridges can be used to hot plug devices.
# Limitations:
# * Currently only pci bridges are supported
# * Until 30 devices per bridge can be hot plugged.
# * Until 5 PCI bridges can be cold plugged per VM.
#   This limitation could be a bug in qemu or in the kernel
# Default number of bridges per SB/VM:
# unspecified or 0   --> will be set to 1
# > 1 <= 5           --> will be set to the specified number
# > 5                --> will be set to 5
default_bridges = 1

# Default memory size in MiB for SB/VM.
# If unspecified then it will be set 2048 MiB.
default_memory = 2048
#
# Default memory slots per SB/VM.
# If unspecified then it will be set 10.
# This is will determine the times that memory will be hotadded to sandbox/VM.
#memory_slots = 10

# The size in MiB will be plused to max memory of hypervisor.
# It is the memory address space for the NVDIMM devie.
# If set block storage driver (block_device_driver) to "nvdimm",
# should set memory_offset to the size of block device.
# Default 0
#memory_offset = 0

# Disable block device from being used for a container's rootfs.
# In case of a storage driver like devicemapper where a container's
# root file system is backed by a block device, the block device is passed
# directly to the hypervisor for performance reasons.
# This flag prevents the block device from being passed to the hypervisor,
# 9pfs is used instead to pass the rootfs.
disable_block_device_use = false

# Block storage driver to be used for the hypervisor in case the container
# rootfs is backed by a block device. This is virtio-scsi, virtio-blk
# or nvdimm.
block_device_driver = "virtio-scsi"

# Specifies cache-related options will be set to block devices or not.
# Default false
#block_device_cache_set = true

# Specifies cache-related options for block devices.
# Denotes whether use of O_DIRECT (bypass the host page cache) is enabled.
# Default false
#block_device_cache_direct = true

# Specifies cache-related options for block devices.
# Denotes whether flush requests for the device are ignored.
# Default false
#block_device_cache_noflush = true

# Enable iothreads (data-plane) to be used. This causes IO to be
# handled in a separate IO thread. This is currently only implemented
# for SCSI.
#
enable_iothreads = false

# Enable pre allocation of VM RAM, default false
# Enabling this will result in lower container density
# as all of the memory will be allocated and locked
# This is useful when you want to reserve all the memory
# upfront or in the cases where you want memory latencies
# to be very predictable
# Default false
#enable_mem_prealloc = true

# Enable huge pages for VM RAM, default false
# Enabling this will result in the VM memory
# being allocated using huge pages.
# This is useful when you want to use vhost-user network
# stacks within the container. This will automatically
# result in memory pre allocation
#enable_hugepages = true

# Enable swap of vm memory. Default false.
# The behaviour is undefined if mem_prealloc is also set to true
#enable_swap = true

# This option changes the default hypervisor and kernel parameters
# to enable debug output where available. This extra output is added
# to the proxy logs, but only when proxy debug is also enabled.
#
# Default false
#enable_debug = true

# Disable the customizations done in the runtime when it detects
# that it is running on top a VMM. This will result in the runtime
# behaving as it would when running on bare metal.
#
#disable_nesting_checks = true

# This is the msize used for 9p shares. It is the number of bytes
# used for 9p packet payload.
#msize_9p = 8192

# If true and vsocks are supported, use vsocks to communicate directly
# with the agent and no proxy is started, otherwise use unix
# sockets and start a proxy to communicate with the agent.
# Default false
#use_vsock = true

# VFIO devices are hotplugged on a bridge by default.
# Enable hotplugging on root bus. This may be required for devices with
# a large PCI bar, as this is a current limitation with hotplugging on
# a bridge. This value is valid for "pc" machine type.
# Default false
#hotplug_vfio_on_root_bus = true

# If host doesn't support vhost_net, set to true. Thus we won't create vhost fds for nics.
# Default false
#disable_vhost_net = true
#
# Default entropy source.
# The path to a host source of entropy (including a real hardware RNG)
# /dev/urandom and /dev/random are two main options.
# Be aware that /dev/random is a blocking source of entropy.  If the host
# runs out of entropy, the VMs boot time will increase leading to get startup
# timeouts.
# The source of entropy /dev/urandom is non-blocking and provides a
# generally acceptable source of entropy. It should work well for pretty much
# all practical purposes.
#entropy_source= "/dev/urandom"

# Path to OCI hook binaries in the *guest rootfs*.
# This does not affect host-side hooks which must instead be added to
# the OCI spec passed to the runtime.
#
# You can create a rootfs with hooks by customizing the osbuilder scripts:
# https://github.com/kata-containers/osbuilder
#
# Hooks must be stored in a subdirectory of guest_hook_path according to their
# hook type, i.e. "guest_hook_path/{prestart,postart,poststop}".
# The agent will scan these directories for executable files and add them, in
# lexicographical order, to the lifecycle of the guest container.
# Hooks are executed in the runtime namespace of the guest. See the official documentation:
# https://github.com/opencontainers/runtime-spec/blob/v1.0.1/config.md#posix-platform-hooks
# Warnings will be logged if any error is encountered will scanning for hooks,
# but it will not abort container execution.
#guest_hook_path = "/usr/share/oci/hooks"

[factory]
# VM templating support. Once enabled, new VMs are created from template
# using vm cloning. They will share the same initial kernel, initramfs and
# agent memory by mapping it readonly. It helps speeding up new container
# creation and saves a lot of memory if there are many kata containers running
# on the same host.
#
# When disabled, new VMs are created from scratch.
#
# Note: Requires "initrd=" to be set ("image=" is not supported).
#
# Default false
#enable_template = true

# The number of caches of VMCache:
# unspecified or == 0   --> VMCache is disabled
# > 0                   --> will be set to the specified number
#
# VMCache is a function that creates VMs as caches before using it.
# It helps speed up new container creation.
# The function consists of a server and some clients communicating
# through Unix socket.  The protocol is gRPC in protocols/cache/cache.proto.
# The VMCache server will create some VMs and cache them by factory cache.
# It will convert the VM to gRPC format and transport it when gets
# requestion from clients.
# Factory grpccache is the VMCache client.  It will request gRPC format
# VM and convert it back to a VM.  If VMCache function is enabled,
# kata-runtime will request VM from factory grpccache when it creates
# a new sandbox.
#
# Default 0
#vm_cache_number = 0

# Specify the address of the Unix socket that is used by VMCache.
#
# Default /var/run/kata-containers/cache.sock
#vm_cache_endpoint = "/var/run/kata-containers/cache.sock"

[proxy.kata]
path = "/opt/kata/libexec/kata-containers/kata-proxy"

# If enabled, proxy messages will be sent to the system log
# (default: disabled)
#enable_debug = true

[shim.kata]
path = "/opt/kata/libexec/kata-containers/kata-shim"

# If enabled, shim messages will be sent to the system log
# (default: disabled)
#enable_debug = true

# If enabled, the shim will create opentracing.io traces and spans.
# (See https://www.jaegertracing.io/docs/getting-started).
#
# Note: By default, the shim runs in a separate network namespace. Therefore,
# to allow it to send trace details to the Jaeger agent running on the host,
# it is necessary to set 'disable_new_netns=true' so that it runs in the host
# network namespace.
#
# (default: disabled)
#enable_tracing = true

[agent.kata]
# There is no field for this section. The goal is only to be able to
# specify which type of agent the user wants to use.

[netmon]
# If enabled, the network monitoring process gets started when the
# sandbox is created. This allows for the detection of some additional
# network being added to the existing network namespace, after the
# sandbox has been created.
# (default: disabled)
#enable_netmon = true

# Specify the path to the netmon binary.
path = "/opt/kata/libexec/kata-containers/kata-netmon"

# If enabled, netmon messages will be sent to the system log
# (default: disabled)
#enable_debug = true

[runtime]
# If enabled, the runtime will log additional debug messages to the
# system log
# (default: disabled)
#enable_debug = true
#
# Internetworking model
# Determines how the VM should be connected to the
# the container network interface
# Options:
#
#   - bridged
#     Uses a linux bridge to interconnect the container interface to
#     the VM. Works for most cases except macvlan and ipvlan.
#
#   - macvtap
#     Used when the Container network interface can be bridged using
#     macvtap.
#
#   - none
#     Used when customize network. Only creates a tap device. No veth pair.
#
#   - tcfilter
#     Uses tc filter rules to redirect traffic from the network interface
#     provided by plugin to a tap interface connected to the VM.
#
internetworking_model="macvtap"

# disable guest seccomp
# Determines whether container seccomp profiles are passed to the virtual
# machine and applied by the kata agent. If set to true, seccomp is not applied
# within the guest
# (default: true)
disable_guest_seccomp=true

# If enabled, the runtime will create opentracing.io traces and spans.
# (See https://www.jaegertracing.io/docs/getting-started).
# (default: disabled)
#enable_tracing = true

# If enabled, the runtime will not create a network namespace for shim and hypervisor processes.
# This option may have some potential impacts to your host. It should only be used when you know what you're doing.
# `disable_new_netns` conflicts with `enable_netmon`
# `disable_new_netns` conflicts with `internetworking_model=bridged` and `internetworking_model=macvtap`. It works only
# with `internetworking_model=none`. The tap device will be in the host network namespace and can connect to a bridge
# (like OVS) directly.
# If you are using docker, `disable_new_netns` only works with `docker run --net=none`
# (default: false)
#disable_new_netns = true

Output of "cat "/opt/kata/share/defaults/kata-containers/configuration.toml"":

# Copyright (c) 2017-2019 Intel Corporation
#
# SPDX-License-Identifier: Apache-2.0
#

# XXX: WARNING: this file is auto-generated.
# XXX:
# XXX: Source file: "cli/config/configuration-qemu.toml.in"
# XXX: Project:
# XXX:   Name: Kata Containers
# XXX:   Type: kata

[hypervisor.qemu]
path = "/opt/kata/bin/qemu-system-x86_64"
kernel = "/opt/kata/share/kata-containers/vmlinuz.container"
image = "/opt/kata/share/kata-containers/kata-containers.img"
machine_type = "pc"

# Optional space-separated list of options to pass to the guest kernel.
# For example, use `kernel_params = "vsyscall=emulate"` if you are having
# trouble running pre-2.15 glibc.
#
# WARNING: - any parameter specified here will take priority over the default
# parameter value of the same name used to start the virtual machine.
# Do not set values here unless you understand the impact of doing so as you
# may stop the virtual machine from booting.
# To see the list of default parameters, enable hypervisor debug, create a
# container and look for 'default-kernel-parameters' log entries.
kernel_params = ""

# Path to the firmware.
# If you want that qemu uses the default firmware leave this option empty
firmware = ""

# Machine accelerators
# comma-separated list of machine accelerators to pass to the hypervisor.
# For example, `machine_accelerators = "nosmm,nosmbus,nosata,nopit,static-prt,nofw"`
machine_accelerators=""

# Default number of vCPUs per SB/VM:
# unspecified or 0                --> will be set to 1
# < 0                             --> will be set to the actual number of physical cores
# > 0 <= number of physical cores --> will be set to the specified number
# > number of physical cores      --> will be set to the actual number of physical cores
default_vcpus = 1

# Default maximum number of vCPUs per SB/VM:
# unspecified or == 0             --> will be set to the actual number of physical cores or to the maximum number
#                                     of vCPUs supported by KVM if that number is exceeded
# > 0 <= number of physical cores --> will be set to the specified number
# > number of physical cores      --> will be set to the actual number of physical cores or to the maximum number
#                                     of vCPUs supported by KVM if that number is exceeded
# WARNING: Depending of the architecture, the maximum number of vCPUs supported by KVM is used when
# the actual number of physical cores is greater than it.
# WARNING: Be aware that this value impacts the virtual machine's memory footprint and CPU
# the hotplug functionality. For example, `default_maxvcpus = 240` specifies that until 240 vCPUs
# can be added to a SB/VM, but the memory footprint will be big. Another example, with
# `default_maxvcpus = 8` the memory footprint will be small, but 8 will be the maximum number of
# vCPUs supported by the SB/VM. In general, we recommend that you do not edit this variable,
# unless you know what are you doing.
default_maxvcpus = 0

# Bridges can be used to hot plug devices.
# Limitations:
# * Currently only pci bridges are supported
# * Until 30 devices per bridge can be hot plugged.
# * Until 5 PCI bridges can be cold plugged per VM.
#   This limitation could be a bug in qemu or in the kernel
# Default number of bridges per SB/VM:
# unspecified or 0   --> will be set to 1
# > 1 <= 5           --> will be set to the specified number
# > 5                --> will be set to 5
default_bridges = 1

# Default memory size in MiB for SB/VM.
# If unspecified then it will be set 2048 MiB.
default_memory = 2048
#
# Default memory slots per SB/VM.
# If unspecified then it will be set 10.
# This is will determine the times that memory will be hotadded to sandbox/VM.
#memory_slots = 10

# The size in MiB will be plused to max memory of hypervisor.
# It is the memory address space for the NVDIMM devie.
# If set block storage driver (block_device_driver) to "nvdimm",
# should set memory_offset to the size of block device.
# Default 0
#memory_offset = 0

# Disable block device from being used for a container's rootfs.
# In case of a storage driver like devicemapper where a container's
# root file system is backed by a block device, the block device is passed
# directly to the hypervisor for performance reasons.
# This flag prevents the block device from being passed to the hypervisor,
# 9pfs is used instead to pass the rootfs.
disable_block_device_use = false

# Block storage driver to be used for the hypervisor in case the container
# rootfs is backed by a block device. This is virtio-scsi, virtio-blk
# or nvdimm.
block_device_driver = "virtio-scsi"

# Specifies cache-related options will be set to block devices or not.
# Default false
#block_device_cache_set = true

# Specifies cache-related options for block devices.
# Denotes whether use of O_DIRECT (bypass the host page cache) is enabled.
# Default false
#block_device_cache_direct = true

# Specifies cache-related options for block devices.
# Denotes whether flush requests for the device are ignored.
# Default false
#block_device_cache_noflush = true

# Enable iothreads (data-plane) to be used. This causes IO to be
# handled in a separate IO thread. This is currently only implemented
# for SCSI.
#
enable_iothreads = false

# Enable pre allocation of VM RAM, default false
# Enabling this will result in lower container density
# as all of the memory will be allocated and locked
# This is useful when you want to reserve all the memory
# upfront or in the cases where you want memory latencies
# to be very predictable
# Default false
#enable_mem_prealloc = true

# Enable huge pages for VM RAM, default false
# Enabling this will result in the VM memory
# being allocated using huge pages.
# This is useful when you want to use vhost-user network
# stacks within the container. This will automatically
# result in memory pre allocation
#enable_hugepages = true

# Enable swap of vm memory. Default false.
# The behaviour is undefined if mem_prealloc is also set to true
#enable_swap = true

# This option changes the default hypervisor and kernel parameters
# to enable debug output where available. This extra output is added
# to the proxy logs, but only when proxy debug is also enabled.
#
# Default false
#enable_debug = true

# Disable the customizations done in the runtime when it detects
# that it is running on top a VMM. This will result in the runtime
# behaving as it would when running on bare metal.
#
#disable_nesting_checks = true

# This is the msize used for 9p shares. It is the number of bytes
# used for 9p packet payload.
#msize_9p = 8192

# If true and vsocks are supported, use vsocks to communicate directly
# with the agent and no proxy is started, otherwise use unix
# sockets and start a proxy to communicate with the agent.
# Default false
#use_vsock = true

# VFIO devices are hotplugged on a bridge by default.
# Enable hotplugging on root bus. This may be required for devices with
# a large PCI bar, as this is a current limitation with hotplugging on
# a bridge. This value is valid for "pc" machine type.
# Default false
#hotplug_vfio_on_root_bus = true

# If host doesn't support vhost_net, set to true. Thus we won't create vhost fds for nics.
# Default false
#disable_vhost_net = true
#
# Default entropy source.
# The path to a host source of entropy (including a real hardware RNG)
# /dev/urandom and /dev/random are two main options.
# Be aware that /dev/random is a blocking source of entropy.  If the host
# runs out of entropy, the VMs boot time will increase leading to get startup
# timeouts.
# The source of entropy /dev/urandom is non-blocking and provides a
# generally acceptable source of entropy. It should work well for pretty much
# all practical purposes.
#entropy_source= "/dev/urandom"

# Path to OCI hook binaries in the *guest rootfs*.
# This does not affect host-side hooks which must instead be added to
# the OCI spec passed to the runtime.
#
# You can create a rootfs with hooks by customizing the osbuilder scripts:
# https://github.com/kata-containers/osbuilder
#
# Hooks must be stored in a subdirectory of guest_hook_path according to their
# hook type, i.e. "guest_hook_path/{prestart,postart,poststop}".
# The agent will scan these directories for executable files and add them, in
# lexicographical order, to the lifecycle of the guest container.
# Hooks are executed in the runtime namespace of the guest. See the official documentation:
# https://github.com/opencontainers/runtime-spec/blob/v1.0.1/config.md#posix-platform-hooks
# Warnings will be logged if any error is encountered will scanning for hooks,
# but it will not abort container execution.
#guest_hook_path = "/usr/share/oci/hooks"

[factory]
# VM templating support. Once enabled, new VMs are created from template
# using vm cloning. They will share the same initial kernel, initramfs and
# agent memory by mapping it readonly. It helps speeding up new container
# creation and saves a lot of memory if there are many kata containers running
# on the same host.
#
# When disabled, new VMs are created from scratch.
#
# Note: Requires "initrd=" to be set ("image=" is not supported).
#
# Default false
#enable_template = true

# The number of caches of VMCache:
# unspecified or == 0   --> VMCache is disabled
# > 0                   --> will be set to the specified number
#
# VMCache is a function that creates VMs as caches before using it.
# It helps speed up new container creation.
# The function consists of a server and some clients communicating
# through Unix socket.  The protocol is gRPC in protocols/cache/cache.proto.
# The VMCache server will create some VMs and cache them by factory cache.
# It will convert the VM to gRPC format and transport it when gets
# requestion from clients.
# Factory grpccache is the VMCache client.  It will request gRPC format
# VM and convert it back to a VM.  If VMCache function is enabled,
# kata-runtime will request VM from factory grpccache when it creates
# a new sandbox.
#
# Default 0
#vm_cache_number = 0

# Specify the address of the Unix socket that is used by VMCache.
#
# Default /var/run/kata-containers/cache.sock
#vm_cache_endpoint = "/var/run/kata-containers/cache.sock"

[proxy.kata]
path = "/opt/kata/libexec/kata-containers/kata-proxy"

# If enabled, proxy messages will be sent to the system log
# (default: disabled)
#enable_debug = true

[shim.kata]
path = "/opt/kata/libexec/kata-containers/kata-shim"

# If enabled, shim messages will be sent to the system log
# (default: disabled)
#enable_debug = true

# If enabled, the shim will create opentracing.io traces and spans.
# (See https://www.jaegertracing.io/docs/getting-started).
#
# Note: By default, the shim runs in a separate network namespace. Therefore,
# to allow it to send trace details to the Jaeger agent running on the host,
# it is necessary to set 'disable_new_netns=true' so that it runs in the host
# network namespace.
#
# (default: disabled)
#enable_tracing = true

[agent.kata]
# There is no field for this section. The goal is only to be able to
# specify which type of agent the user wants to use.

[netmon]
# If enabled, the network monitoring process gets started when the
# sandbox is created. This allows for the detection of some additional
# network being added to the existing network namespace, after the
# sandbox has been created.
# (default: disabled)
#enable_netmon = true

# Specify the path to the netmon binary.
path = "/opt/kata/libexec/kata-containers/kata-netmon"

# If enabled, netmon messages will be sent to the system log
# (default: disabled)
#enable_debug = true

[runtime]
# If enabled, the runtime will log additional debug messages to the
# system log
# (default: disabled)
#enable_debug = true
#
# Internetworking model
# Determines how the VM should be connected to the
# the container network interface
# Options:
#
#   - bridged
#     Uses a linux bridge to interconnect the container interface to
#     the VM. Works for most cases except macvlan and ipvlan.
#
#   - macvtap
#     Used when the Container network interface can be bridged using
#     macvtap.
#
#   - none
#     Used when customize network. Only creates a tap device. No veth pair.
#
#   - tcfilter
#     Uses tc filter rules to redirect traffic from the network interface
#     provided by plugin to a tap interface connected to the VM.
#
internetworking_model="macvtap"

# disable guest seccomp
# Determines whether container seccomp profiles are passed to the virtual
# machine and applied by the kata agent. If set to true, seccomp is not applied
# within the guest
# (default: true)
disable_guest_seccomp=true

# If enabled, the runtime will create opentracing.io traces and spans.
# (See https://www.jaegertracing.io/docs/getting-started).
# (default: disabled)
#enable_tracing = true

# If enabled, the runtime will not create a network namespace for shim and hypervisor processes.
# This option may have some potential impacts to your host. It should only be used when you know what you're doing.
# `disable_new_netns` conflicts with `enable_netmon`
# `disable_new_netns` conflicts with `internetworking_model=bridged` and `internetworking_model=macvtap`. It works only
# with `internetworking_model=none`. The tap device will be in the host network namespace and can connect to a bridge
# (like OVS) directly.
# If you are using docker, `disable_new_netns` only works with `docker run --net=none`
# (default: false)
#disable_new_netns = true

Config file /usr/share/defaults/kata-containers/configuration.toml not found

---

KSM throttler

version

Output of " --version":

./kata-collect-data.sh: line 175: --version: command not found

systemd service

Image details

---
osbuilder:
  url: "https://github.com/kata-containers/osbuilder"
  version: "1.6.0-e888ef5a601bf5058937bebab48a275333eb4566"
rootfs-creation-time: "2019-03-26T23:54:37.920591376+0000Z"
description: "osbuilder rootfs"
file-format-version: "0.0.2"
architecture: "x86_64"
base-distro:
  name: "bionic"
  version: "18.04"
  packages:
    default:
      - "systemd,iptables,init,chrony,bash"
    extra:
      - "bash"
agent:
  url: "https://github.com/kata-containers/agent"
  name: "kata-agent"
  version: "1.6.0-rc2-48dd1c031530fce9bf16b0f6a7305979cedd8fc9"
  agent-is-init-daemon: "no"

---

Initrd details

No initrd

---

Logfiles

Runtime logs

No recent runtime problems found in system journal.

Proxy logs

No recent proxy problems found in system journal.

Shim logs

No recent shim problems found in system journal.

Throttler logs

No recent throttler problems found in system journal.

---

Container manager details

Have docker

Docker

Output of "docker version":

Client:
 Version:           18.06.1-ce
 API version:       1.38
 Go version:        go1.10.8
 Git commit:        e68fc7a
 Built:             Tue Aug 21 17:16:31 2018
 OS/Arch:           linux/amd64
 Experimental:      false
Cannot connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?

Output of "docker info":

Cannot connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?

Output of "systemctl show docker":

Restart=no
NotifyAccess=none
RestartUSec=100ms
TimeoutStartUSec=1min 30s
TimeoutStopUSec=1min 30s
RuntimeMaxUSec=infinity
WatchdogUSec=0
WatchdogTimestampMonotonic=0
PermissionsStartOnly=no
RootDirectoryStartOnly=no
RemainAfterExit=no
GuessMainPID=yes
MainPID=0
ControlPID=0
FileDescriptorStoreMax=0
NFileDescriptorStore=0
StatusErrno=0
Result=success
UID=[not set]
GID=[not set]
NRestarts=0
ExecMainStartTimestampMonotonic=0
ExecMainExitTimestampMonotonic=0
ExecMainPID=0
ExecMainCode=0
ExecMainStatus=0
MemoryCurrent=[not set]
CPUUsageNSec=[not set]
TasksCurrent=[not set]
IPIngressBytes=18446744073709551615
IPIngressPackets=18446744073709551615
IPEgressBytes=18446744073709551615
IPEgressPackets=18446744073709551615
Delegate=no
CPUAccounting=no
CPUWeight=[not set]
StartupCPUWeight=[not set]
CPUShares=[not set]
StartupCPUShares=[not set]
CPUQuotaPerSecUSec=infinity
IOAccounting=no
IOWeight=[not set]
StartupIOWeight=[not set]
BlockIOAccounting=no
BlockIOWeight=[not set]
StartupBlockIOWeight=[not set]
MemoryAccounting=yes
MemoryLow=0
MemoryHigh=infinity
MemoryMax=infinity
MemorySwapMax=infinity
MemoryLimit=infinity
DevicePolicy=auto
TasksAccounting=yes
TasksMax=73727
IPAccounting=no
UMask=0022
LimitCPU=infinity
LimitCPUSoft=infinity
LimitFSIZE=infinity
LimitFSIZESoft=infinity
LimitDATA=infinity
LimitDATASoft=infinity
LimitSTACK=infinity
LimitSTACKSoft=8388608
LimitCORE=infinity
LimitCORESoft=infinity
LimitRSS=infinity
LimitRSSSoft=infinity
LimitNOFILE=1048576
LimitNOFILESoft=1048576
LimitAS=infinity
LimitASSoft=infinity
LimitNPROC=2063289
LimitNPROCSoft=2063289
LimitMEMLOCK=16777216
LimitMEMLOCKSoft=16777216
LimitLOCKS=infinity
LimitLOCKSSoft=infinity
LimitSIGPENDING=2063289
LimitSIGPENDINGSoft=2063289
LimitMSGQUEUE=819200
LimitMSGQUEUESoft=819200
LimitNICE=0
LimitNICESoft=0
LimitRTPRIO=0
LimitRTPRIOSoft=0
LimitRTTIME=infinity
LimitRTTIMESoft=infinity
OOMScoreAdjust=0
Nice=0
IOSchedulingClass=0
IOSchedulingPriority=0
CPUSchedulingPolicy=0
CPUSchedulingPriority=0
TimerSlackNSec=50000
CPUSchedulingResetOnFork=no
NonBlocking=no
StandardInput=null
StandardInputData=
StandardOutput=inherit
StandardError=inherit
TTYReset=no
TTYVHangup=no
TTYVTDisallocate=no
SyslogPriority=30
SyslogLevelPrefix=yes
SyslogLevel=6
SyslogFacility=3
LogLevelMax=-1
SecureBits=0
CapabilityBoundingSet=cap_chown cap_dac_override cap_dac_read_search cap_fowner cap_fsetid cap_kill cap_setgid cap_setuid cap_setpcap cap_linux_immutable cap_net_bind_service cap_net_broadcast cap_net_admin cap_net_raw cap_ipc_lock cap_ipc_owner cap_sys_module cap_sys_rawio cap_sys_chroot cap_sys_ptrace cap_sys_pacct cap_sys_admin cap_sys_boot cap_sys_nice cap_sys_resource cap_sys_time cap_sys_tty_config cap_mknod cap_lease cap_audit_write cap_audit_control cap_setfcap cap_mac_override cap_mac_admin cap_syslog cap_wake_alarm cap_block_suspend
AmbientCapabilities=
DynamicUser=no
RemoveIPC=no
MountFlags=
PrivateTmp=no
PrivateDevices=no
ProtectKernelTunables=no
ProtectKernelModules=no
ProtectControlGroups=no
PrivateNetwork=no
PrivateUsers=no
ProtectHome=no
ProtectSystem=no
SameProcessGroup=no
UtmpMode=init
IgnoreSIGPIPE=yes
NoNewPrivileges=no
SystemCallErrorNumber=0
LockPersonality=no
RuntimeDirectoryPreserve=no
RuntimeDirectoryMode=0755
StateDirectoryMode=0755
CacheDirectoryMode=0755
LogsDirectoryMode=0755
ConfigurationDirectoryMode=0755
MemoryDenyWriteExecute=no
RestrictRealtime=no
RestrictNamespaces=no
MountAPIVFS=no
KeyringMode=private
KillMode=control-group
KillSignal=15
SendSIGKILL=yes
SendSIGHUP=no
Id=docker.service
Names=docker.service
WantedBy=kitt-init.service
ConsistsOf=docker.socket
Before=kitt-init.service
After=docker.socket
TriggeredBy=docker.socket
Description=docker.service
LoadState=masked
ActiveState=inactive
SubState=dead
FragmentPath=/dev/null
UnitFileState=masked
StateChangeTimestamp=Tue 2019-03-26 23:22:33 UTC
StateChangeTimestampMonotonic=388268429
InactiveExitTimestampMonotonic=0
ActiveEnterTimestampMonotonic=0
ActiveExitTimestampMonotonic=0
InactiveEnterTimestampMonotonic=0
CanStart=no
CanStop=yes
CanReload=no
CanIsolate=no
StopWhenUnneeded=no
RefuseManualStart=no
RefuseManualStop=no
AllowIsolate=no
DefaultDependencies=yes
OnFailureJobMode=replace
IgnoreOnIsolate=no
NeedDaemonReload=no
JobTimeoutUSec=infinity
JobRunningTimeoutUSec=infinity
JobTimeoutAction=none
ConditionResult=no
AssertResult=no
ConditionTimestampMonotonic=0
AssertTimestampMonotonic=0
Transient=no
Perpetual=no
StartLimitIntervalUSec=10s
StartLimitBurst=5
StartLimitAction=none
FailureAction=none
SuccessAction=none
CollectMode=inactive

No kubectl
No crio
Have containerd

containerd

Output of "containerd --version":

containerd github.com/containerd/containerd 1.1.2 468a545b9edcd5932818eb9de8e72413e616e86e

Output of "systemctl show containerd":

Type=simple
Restart=always
NotifyAccess=none
RestartUSec=5s
TimeoutStartUSec=1min 30s
TimeoutStopUSec=1min 30s
RuntimeMaxUSec=infinity
WatchdogUSec=0
WatchdogTimestamp=Tue 2019-03-26 23:24:13 UTC
WatchdogTimestampMonotonic=488528071
PermissionsStartOnly=no
RootDirectoryStartOnly=no
RemainAfterExit=no
GuessMainPID=yes
MainPID=19186
ControlPID=0
FileDescriptorStoreMax=0
NFileDescriptorStore=0
StatusErrno=0
Result=success
UID=[not set]
GID=[not set]
NRestarts=0
ExecMainStartTimestamp=Tue 2019-03-26 23:24:13 UTC
ExecMainStartTimestampMonotonic=488528028
ExecMainExitTimestampMonotonic=0
ExecMainPID=19186
ExecMainCode=0
ExecMainStatus=0
ExecStartPre={ path=/sbin/modprobe ; argv[]=/sbin/modprobe overlay ; ignore_errors=no ; start_time=[Tue 2019-03-26 23:24:11 UTC] ; stop_time=[Tue 2019-03-26 23:24:11 UTC] ; pid=19117 ; code=exited ; status=0 }
ExecStartPre={ path=/opt/bin/containerd-init.sh ; argv[]=/opt/bin/containerd-init.sh ; ignore_errors=no ; start_time=[Tue 2019-03-26 23:24:11 UTC] ; stop_time=[Tue 2019-03-26 23:24:13 UTC] ; pid=19118 ; code=exited ; status=0 }
ExecStart={ path=/opt/containerd/bin/containerd ; argv[]=/opt/containerd/bin/containerd --log-level=info --config=/etc/containerd/config.toml ; ignore_errors=no ; start_time=[Tue 2019-03-26 23:24:13 UTC] ; stop_time=[n/a] ; pid=19186 ; code=(null) ; status=0/0 }
Slice=system.slice
ControlGroup=/system.slice/containerd.service
MemoryCurrent=3669561344
CPUUsageNSec=[not set]
TasksCurrent=471
IPIngressBytes=18446744073709551615
IPIngressPackets=18446744073709551615
IPEgressBytes=18446744073709551615
IPEgressPackets=18446744073709551615
Delegate=yes
DelegateControllers=cpu cpuacct io blkio memory devices pids
CPUAccounting=no
CPUWeight=[not set]
StartupCPUWeight=[not set]
CPUShares=[not set]
StartupCPUShares=[not set]
CPUQuotaPerSecUSec=infinity
IOAccounting=no
IOWeight=[not set]
StartupIOWeight=[not set]
BlockIOAccounting=no
BlockIOWeight=[not set]
StartupBlockIOWeight=[not set]
MemoryAccounting=yes
MemoryLow=0
MemoryHigh=infinity
MemoryMax=infinity
MemorySwapMax=infinity
MemoryLimit=infinity
DevicePolicy=auto
TasksAccounting=yes
TasksMax=73727
IPAccounting=no
UMask=0022
LimitCPU=infinity
LimitCPUSoft=infinity
LimitFSIZE=infinity
LimitFSIZESoft=infinity
LimitDATA=infinity
LimitDATASoft=infinity
LimitSTACK=infinity
LimitSTACKSoft=8388608
LimitCORE=infinity
LimitCORESoft=infinity
LimitRSS=infinity
LimitRSSSoft=infinity
LimitNOFILE=1048576
LimitNOFILESoft=1048576
LimitAS=infinity
LimitASSoft=infinity
LimitNPROC=infinity
LimitNPROCSoft=infinity
LimitMEMLOCK=16777216
LimitMEMLOCKSoft=16777216
LimitLOCKS=infinity
LimitLOCKSSoft=infinity
LimitSIGPENDING=2063289
LimitSIGPENDINGSoft=2063289
LimitMSGQUEUE=819200
LimitMSGQUEUESoft=819200
LimitNICE=0
LimitNICESoft=0
LimitRTPRIO=0
LimitRTPRIOSoft=0
LimitRTTIME=infinity
LimitRTTIMESoft=infinity
OOMScoreAdjust=-999
Nice=0
IOSchedulingClass=0
IOSchedulingPriority=0
CPUSchedulingPolicy=0
CPUSchedulingPriority=0
TimerSlackNSec=50000
CPUSchedulingResetOnFork=no
NonBlocking=no
StandardInput=null
StandardInputData=
StandardOutput=journal
StandardError=inherit
TTYReset=no
TTYVHangup=no
TTYVTDisallocate=no
SyslogPriority=30
SyslogLevelPrefix=yes
SyslogLevel=6
SyslogFacility=3
LogLevelMax=-1
SecureBits=0
CapabilityBoundingSet=cap_chown cap_dac_override cap_dac_read_search cap_fowner cap_fsetid cap_kill cap_setgid cap_setuid cap_setpcap cap_linux_immutable cap_net_bind_service cap_net_broadcast cap_net_admin cap_net_raw cap_ipc_lock cap_ipc_owner cap_sys_module cap_sys_rawio cap_sys_chroot cap_sys_ptrace cap_sys_pacct cap_sys_admin cap_sys_boot cap_sys_nice cap_sys_resource cap_sys_time cap_sys_tty_config cap_mknod cap_lease cap_audit_write cap_audit_control cap_setfcap cap_mac_override cap_mac_admin cap_syslog cap_wake_alarm cap_block_suspend
AmbientCapabilities=
DynamicUser=no
RemoveIPC=no
MountFlags=
PrivateTmp=no
PrivateDevices=no
ProtectKernelTunables=no
ProtectKernelModules=no
ProtectControlGroups=no
PrivateNetwork=no
PrivateUsers=no
ProtectHome=no
ProtectSystem=no
SameProcessGroup=no
UtmpMode=init
IgnoreSIGPIPE=yes
NoNewPrivileges=no
SystemCallErrorNumber=0
LockPersonality=no
RuntimeDirectoryPreserve=no
RuntimeDirectoryMode=0755
StateDirectoryMode=0755
CacheDirectoryMode=0755
LogsDirectoryMode=0755
ConfigurationDirectoryMode=0755
MemoryDenyWriteExecute=no
RestrictRealtime=no
RestrictNamespaces=no
MountAPIVFS=no
KeyringMode=private
KillMode=process
KillSignal=15
SendSIGKILL=yes
SendSIGHUP=no
Id=containerd.service
Names=containerd.service
Requires=sysinit.target system.slice
WantedBy=multi-user.target
Conflicts=shutdown.target
Before=shutdown.target multi-user.target
After=sysinit.target kata-init.service containerd-devicemapper-init.service basic.target system.slice systemd-journald.socket
Documentation=https://containerd.io
Description=containerd container runtime
LoadState=loaded
ActiveState=active
SubState=running
FragmentPath=/etc/systemd/system/containerd.service
UnitFileState=enabled
UnitFilePreset=enabled
StateChangeTimestamp=Tue 2019-03-26 23:24:13 UTC
StateChangeTimestampMonotonic=488528073
InactiveExitTimestamp=Tue 2019-03-26 23:24:11 UTC
InactiveExitTimestampMonotonic=486885031
ActiveEnterTimestamp=Tue 2019-03-26 23:24:13 UTC
ActiveEnterTimestampMonotonic=488528073
ActiveExitTimestamp=Tue 2019-03-26 23:22:16 UTC
ActiveExitTimestampMonotonic=371661373
InactiveEnterTimestamp=Tue 2019-03-26 23:22:16 UTC
InactiveEnterTimestampMonotonic=371670496
CanStart=yes
CanStop=yes
CanReload=no
CanIsolate=no
StopWhenUnneeded=no
RefuseManualStart=no
RefuseManualStop=no
AllowIsolate=no
DefaultDependencies=yes
OnFailureJobMode=replace
IgnoreOnIsolate=no
NeedDaemonReload=no
JobTimeoutUSec=infinity
JobRunningTimeoutUSec=infinity
JobTimeoutAction=none
ConditionResult=yes
AssertResult=yes
ConditionTimestamp=Tue 2019-03-26 23:24:11 UTC
ConditionTimestampMonotonic=486883363
AssertTimestamp=Tue 2019-03-26 23:24:11 UTC
AssertTimestampMonotonic=486883364
Transient=no
Perpetual=no
StartLimitIntervalUSec=10s
StartLimitBurst=5
StartLimitAction=none
FailureAction=none
SuccessAction=none
InvocationID=7363f32d088f436b9c8704a3af44b18a
CollectMode=inactive

Output of "cat /etc/containerd/config.toml":

[grpc]
address = "/run/containerd/containerd.sock"
uid = 0
gid = 0

[plugins]
  [plugins.devmapper]
    pool_name = "containerd-pool"
    base_image_size = "32GB"
  [plugins.cri.containerd]
    snapshotter = "devmapper"
    [plugins.cri.containerd.default_runtime]
      runtime_type = "io.containerd.runtime.v1.linux"
      runtime_engine = "/usr/bin/runc"
      runtime_root = ""
    [plugins.cri.containerd.untrusted_workload_runtime]
      runtime_type = "io.containerd.kata.v2"

[plugins.cri]
  max_container_log_line_size = 262144

[plugins.linux]
  shim = "/opt/containerd/bin/containerd-shim"
  runtime = "runc"

[plugins.cri.registry]
  [plugins.cri.registry.mirrors]
    [plugins.cri.registry.mirrors."docker.io"]
       endpoint = []

---

Packages

No dpkg
No rpm

---

[devimc]

@cespo It's pretty slow because of 9p, this issues will be fixed once kata supports virtio-fs. see #1016

[egernst]

@devimc they are using devicemapper

[devimc]

@egernst ok, then why they are sharing /var/run/ ?

[egernst]

At least we should a look and see why the "DinK" (docker in kata) setup has Native Overlay Diff disabled, then.

They are sharing so they can utilize the docker socket (they are doing docker-in-docker). I need to better understand the flow for DiD wrt what is done with the docker socket from the host, but I'm assuming build context is moving from guest -> host.

If this is the case, we could consider an image that isn't exactly DinD, but instead Kata specific to use its own daemon instead of hosts. I'm not sure if the usage with runc leverages a build cache, or something, though...

[egernst]

When performance is terrible, always look at 9p, even if you don't think you're using it.

I could reproduce it, and have a suggested fix. After 8 minutes of still extracting, I gave up running a test docker image. With the fix, it finishes the build in ~2 minutes

A few items:

The fix

Not particularly relevant, but here's the docker image I was testing to build inside DinD:

FROM nefelinetworks/bess_build
ARG BESS_COMMIT=master
RUN apt-get update && apt-get install -y wget unzip ca-certificates git
RUN wget -qO bess.zip https://github.com/NetSys/bess/archive/${BESS_COMMIT}.zip && unzip bess.zip
WORKDIR bess-${BESS_COMMIT}

Here's the "original" yaml for dind (though I updated memory requirements for the pod, and utilize runtimeClass):

apiVersion: v1
kind: Pod
metadata:
  name: dind-terrible
spec:
  runtimeClassName: kata
  containers:
  - name: dind
    securityContext:
      privileged: true
    image: docker:18.06-dind
    args: ["--storage-driver=overlay2"]
    resources:
      limits:
        memory: "3G"
    volumeMounts:
      - mountPath: /var/run/
        name: dockersock
  volumes:
    - name: dockersock
      emptyDir: {}

With this, when exec'd in, we see:

/ # mount | grep lib\/docker
kataShared on /var/lib/docker type 9p (rw,dirsync,nodev,relatime,mmap,access=client,trans=virtio)

Terrible, terrible. By simply creating the volume explicitly for /var/lib/docker, and back it by a tmpfs, we avoid the issue. Updated yaml:

apiVersion: v1
kind: Pod
metadata:
  name: dind
spec:
  runtimeClassName: kata
  containers:
  - name: dind
    securityContext:
      privileged: true
    image: docker:18.06-dind
    args: ["--storage-driver=overlay2"]
    resources:
      limits:
        memory: "3G"
    volumeMounts:
      - mountPath: /var/run/
        name: dockersock
      - mountPath: /var/lib/docker
        name: docker
  volumes:
    - name: dockersock
      emptyDir: {}
    - name: docker
      emptyDir:
        medium: Memory

When exec'd in we see:

/ # mount | grep lib\/docker
tmpfs on /var/lib/docker type tmpfs (rw,relatime)

With this, /var/lib/docker is handled by a tmpfs instead of 9p. The tradeoff, now, is disk-space for memory, since the image building itself will be backed by RAM. In the example above, we specified 3GB memory for the pod to handle this overhead.

Other considerations:

When running your yaml as is, you are running unconstrained. So, if you do it with runc on a 96 core machine, your build and extractions will be parallel. With Kata default, you'll only run with a single CPU. Similar factors should be considered with respect to RAM. From discussions in the past, I understand that you have constraints dialed in pretty well, so I'm sure you're already very aware!

Thanks to @mcastelino and @krsna1729 for the fun debug session!

[mcastelino]

@cespo for a little more context you are trading memory for storage

https://kubernetes.io/docs/concepts/storage/volumes/#emptydir

By default, emptyDir volumes are stored on whatever medium is backing the node - that might be disk or SSD or network storage, depending on your environment. However, you can set the emptyDir.medium field to "Memory" to tell Kubernetes to mount a tmpfs (RAM-backed filesystem) for you instead. While tmpfs is very fast, be aware that unlike disks, tmpfs is cleared on node reboot and any files you write will count against your Container’s memory limit.

If you do not want to trade memory for storage or want to persist you may want to explore using raw block volumes. However in this case as you really want an ephemeral build using memory may not be too bad.

https://kubernetes.io/blog/2019/03/07/raw-block-volume-support-to-beta/

We have verified that it works with kata
https://gist.github.com/amshinde/2ab9b5a2f2e91694a7421b1e2b787b58

/cc @amshinde

[mcastelino]

Also once we add virtio-fs support via nemu to kata you no longer need to make this tradeoff.

[cespo]

We don't really have to trade memory for storage in this case.
The problem with the dind image is that a volume is created during the build of the image (https://github.com/docker-library/docker/blob/65fab2cd767c10f22ee66afa919eda80dbdc8872/18.09/dind/Dockerfile) and it gets mounted as kataShared filesystem.
Commenting that out and rebuilding the image moved /var/lib/docker to the disk which is on devicemapper.

The problem with 9pfs still stands here though because everything that is shared between containers is still super slow. Looking forward to seeing virtio-fs support landing in a official release.
Any ETA?

[sboeuf]

@cespo

Looking forward to seeing virtio-fs support landing in a official release.
Any ETA?

The target for an early support (experimental since not landed upstream QEMU/kernel yet) is for next Kata release 1.7 (mid May).

[mcastelino]

Commenting that out and rebuilding the image moved /var/lib/docker to the disk which is on devicemapper.

@cespo commenting that volume out will help solve the issue.

However you will still be bound the size of the max size of devicemapper, which by default is configured to be 10GB. So if the build results in increasing the container growing to that limit, you may see failures. And this limit is a node level configuration (at least with docker it was).

Just something to watch for.

[egernst]

Hey @cespo - can this issue be closed, or are there still opens?