This example reproduces the issue when using LXC together with Log4Cplus logging library.
It is related to fork system call in do_lxcapi_create call inside liblxc.
For LXC 2.0.8: lxccontainer.c:1505
/*
* Create the backing store
* Note we can't do this in the same task as we use to execute the
* template because of the way zfs works.
* After you 'zfs create', zfs mounts the fs only in the initial
* namespace.
*/
pid = fork();
if (pid < 0) {
SYSERROR("failed to fork task for container creation template");
goto out_unlock;
}
if (pid == 0) { // child
struct bdev *bdev = NULL;
if (!(bdev = do_bdev_create(c, bdevtype, specs))) {
ERROR("Error creating backing store type %s for %s",
bdevtype ? bdevtype : "(none)", c->name);
exit(1);
}
/* save config file again to store the new rootfs location */
if (!do_lxcapi_save_config(c, NULL)) {
ERROR("failed to save starting configuration for %s", c->name);
// parent task won't see bdev in config so we delete it
bdev->ops->umount(bdev);
bdev->ops->destroy(bdev);
exit(1);
}
exit(0);
}
if (wait_for_pid(pid) != 0)
goto out_unlock;
In this configuration, the executable hangs forever after container is created.
Steps:
$ sudo apt install vagrant
$ vagrant up
$ vagrant ssh
In Virtual Machine:
cd /vagrant
mkdir build
cd build
cmake ..
sudo TSAN_OPTIONS="halt_on_error=1" ./log4cplus-lxc-issue
sudo valgrind ./log4cplus-lxc-issue