When running in a container, we cannot assume to have access to dmeventd. Up to now, we manually started dmeventd to work around an issue when running the container on systemd with high rlimits. See 2b2ec174bcde for details.
However, there is no need to start dmeventd at all: it's primary use for automatic resizing VGs and LVs doesn't fit the use in containers anyways. So we can just disable "monitoring" of LVs, which in turn should prevent dmeventd from being started by some random lvcreate.
Also, use "exec" to start the linstor processes: that way they get to be PID 1 in the container and they receive any SIGINTs generated by the container runtime.
When running in a container, we cannot assume to have access to dmeventd. Up to now, we manually started dmeventd to work around an issue when running the container on systemd with high rlimits. See 2b2ec174bcde for details.
However, there is no need to start dmeventd at all: it's primary use for automatic resizing VGs and LVs doesn't fit the use in containers anyways. So we can just disable "monitoring" of LVs, which in turn should prevent dmeventd from being started by some random lvcreate.
Also, use "exec" to start the linstor processes: that way they get to be PID 1 in the container and they receive any SIGINTs generated by the container runtime.