This removes the github.com/pkg/errors package in favor of the official
error wrapping in go 1.13.
Signed-off-by: Andrew Rynhard <andrew@andrewrynhard.com>
This moves to using a single revision of pkgs. It includes a few
changes:
- kernel with KVM host support
- containerd v1.3.0
This change brings in a kernel with host KVM support. This will allow us
to use VMs within Talos for things like integrations tests. This also
allows users to do things with KVM as they see fit.
Signed-off-by: Andrew Rynhard <andrew@andrewrynhard.com>
Memory usage reduced around 8-10x: now it stays stable at 1GB.
I disabled some of the new linters, and one rule which is violated a
lot.
I might make sense to go back and enable `wsl` fixing all the issues
(leaving that for another PR).
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
This moves from translating a config into an internal config
representation, to using an interface. The idea is that an interface
gives us stronger compile time checks, and will prevent us from having to copy
from on struct to another. As long as a concrete type implements the
Configurator interface, it can be used to provide instructions to Talos.
Signed-off-by: Andrew Rynhard <andrew@andrewrynhard.com>
Looks like containerd creates shim file sockets in Linux abstract
namespace which are fixed (don't depend on containerd root directory)
and depend on container namespace and id. So if two containerd instances
on the same host run same namespace/id pair, that is going to create a
conflict on that shim filesocket.
Avoid that by randomizing namespace name. CRI tests should be fine as
namespace is fixed, but container ID is random.
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
The gofumpt linter is a stricter drop-in replacement for gofmt. The
rules are ones that I strongly agree with and I think it would be better
if we added this linter instead of nit picking every PR.
Signed-off-by: Andrew Rynhard <andrew@andrewrynhard.com>
It runs containers via CRI interface in a pod sandbox. This is the very
first version: I tried not to introduce any changes to common runner
interface.
There should be some CRI-speficic options for the runner (like polling
interval, as it doesn't have nice `Wait()` API), plus my plan so far is
to use OCI as the common layer for container options, so that we can
analyze OCI and translate to CRI (when possible, return errors when
option is not implemented).
CRI interface doesn't have a concept of 'unpacking' an image, so we
probably need to unpack via containerd API (or any other
runtime-specific API) by targeting CRI namespace.
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
This is preparation for implementing CRI runner.
CRI client moved into its own package, I split it into multiple files
and added rudimentary tests for it.
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
This makes test launch their own isolated instance of containerd with
its own root/state directories and listening socket address. Each test
brings this instance up/down on its own.
Add options to override containerd address in the code (used only in the
tests).
Enable parallel go test runs once again.
P.S. I wish I could share that 'SetupSuite' phase across the tests, but
afaik there's no way in Go to share `_test.go` code across packages. If
we put it as normal package, this might pull in test dependencies (like
`testify`) into production code, which I don't like.
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
This refactors metrics interface to remove containerd-specific stuff and
make it common for CRI & containerd.
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
Instead of pulling a full list of containers, implement inspector query
for a single container following the spec to build display name.
Also adds many more tests for the container inspector.
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
For #711, this should be a complete fix - waiting for container to be
started.
For #712, this should be more of a workaround - playing with timeouts to
hit the failure less likely. Idea of the test is that health check
should be aborted on timeout (1ms) while health check succeeds if not
aborted in 50ms. Before the fix it was 1ms/10ms, but still concurrently
there was a chance that goroutine exits successfully after 10ms while
1ms context deadline is not reached.
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
Fixes: #689, #690
Refactor container inspection code into a package of its own with some
rudimentary tests. Use this package consistently in osd commands dealing
with containers.
Improvements for the next PRs:
* implement API to fetch info about container by ID (to avoid fetching
full list)
* handle and display errors on client side, not to the log of the
server
* more tests, including k8s containers (how can we do that?)
Signed-off-by: Andrey Smirnov <smirnov.andrey@gmail.com>
