We should never return a nil context because of the way this function is
typically used... e.g.
```
ctx, done, err := containerd.WithLease(ctx)
```
If there is an error `ctx` will be nil and any error handling may cause
an NPE if it tries to use `ctx`.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
Currently hardcoded to 2 seconds; in GitHub actions we see random
cancellation of our integration suite right at 2 seconds even
though containerd is within milliseconds of being ready.
Signed-off-by: Phil Estes <estesp@linux.vnet.ibm.com>
TestRuntimeWithEmptyMaxEnvProcs should restore the GoMaxProcs after
test so that the temporary change of GoMaxProcs will not impact other
case, like TestRuntimeWithNonEmptyMaxEnvProcs.
Signed-off-by: Wei Fu <fuweid89@gmail.com>
For some reason, shimv2 process doesn't exist. The ttrpc doesn't detect
the connection closed by server until delete task. For this case, we
should ignore the ttrpc.ErrClosed and let task manager handle the
cleanup.
Signed-off-by: Wei Fu <fuweid89@gmail.com>
containerd loads timeout values from config.toml and populated those
values to `timeout` package at launch. So when using `timeout` package
from shim, there are default values and config file is ignored.
So use a hardcoded value for binary IO.
Signed-off-by: Maksym Pavlenko <makpav@amazon.com>
This was changed to `no_subreaper` in
6e9f24b711 and, as far as I can tell,
`no_subreaper` doesn't exist as a config anymore.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
full diff: 92cb4ed978..61b7af7564
This adds new dependency github.com/fsnotify/fsnotify since 4ce334aa49
Signed-off-by: Kohei Tokunaga <ktokunaga.mail@gmail.com>
This allows Travis to continue running on PRs for release/1.x branches
where we have not enabled GH Actions.
Signed-off-by: Phil Estes <estesp@linux.vnet.ibm.com>
Integration tests were running with latest Go release rather than the
version used everywhere else. Also, we don't need to install protoc from
tarball and also apt-get the package for Ubuntu when used as a
dependency for criu build.
Signed-off-by: Phil Estes <estesp@linux.vnet.ibm.com>
A simple `grep selinux_stub bin/containerd` shows that we are not using
selinux tag when building containerd. The CRI plugin pulls in a bunch of
selinux related code, so we need to enable this tag as well.
Signed-off-by: Davanum Srinivas <davanum@gmail.com>
forkAndMountat forks a process to chdir then mount layers. Signals are
blocked (using runtime_beforeFork) during fork.
There is a race condition that the child process finishes before the
parent process is scheduled and can unblock signal handling. The SIGCHLD
signal sent from the finished process may have been delivered to the
shim process's reaper thread and caused the parent process fail with
ECHLD error.
This patch sets up a pipe for communication between child and parent
instead of waiting for child exit status.
Fixes#4009.
Signed-off-by: Haitao Li <hli@atlassian.com>
Also few changes to names of workflow jobs. Integration tests can get
canceled without full completing at 10 minutes. Make sure we only use
checkout v2 action.
Signed-off-by: Phil Estes <estesp@linux.vnet.ibm.com>
