View this report as an mbox folder.
email@example.com, David Kuehling <firstname.lastname@example.org>, email@example.com:
nbd-client. Full text available.
David Kuehling <firstname.lastname@example.org>:
David Kuehling <email@example.com>, firstname.lastname@example.org. Full text available.
From: David Kuehling <email@example.com> To: Devuan Bug Tracking System <firstname.lastname@example.org> Subject: nbd-client: At shutdown nbd-client disabled before file-systems could be cleanly unmounted Date: Mon, 16 Jul 2018 01:52:51 +0200 Package: nbd-client Version: 1:3.15.2-3 Severity: normal Dear Maintainer, I'm currently running a Devuan desktop without local storage using pxelinux and root located on a network block device provided via nbd-client (kernel command line "root=/dev/mapper/vg-root nbdroot=192.168.0.123,devicename,nbd0"). Unfortunately the sysvinit tools and init scripts and the nbd-client binary do not properly work together causing nbd-client to disconnect at shutdown before the filesystems can be properly unmounted. This seems to be a very old problem documented e.g. here  and here  and it seems that systemd already got specific patches to prevent that problem. With only sysvinit left to exhibit the problem, this sort of makes the bug devuan-specific (or should I still bother to report it to debian?). Short summary of the problem: during shutdown /etc/init.d/sendsigs calls killall5 binary from sysvinit-utils, killing almost all running processes. Of course it never should kill nbd-client, so the /etc/init.d/nbd-client script is smart enough to register its PID to be exempt from sendsig's action: by recording it in the /run/sendsigs.omit.d/nbd-cient file. These PIDs are then collected by /etc/init.d/sendsig and passed as "-o NNN" options to killall5 which spares those processes from premature termination. However, before killall5 goes on to kill all the other processes, it does a: /* Now stop all processes. */ kill(-1, SIGSTOP); And when it's done, it does: /* And let them continue. */ kill(-1, SIGCONT); These SIGSTOP, SIGCONT signals are passed to all processes, including nbd-client. Unfortunately nbd-client is written in a way that makes it unable to handle any signals delivered while it is inside an ioctl call, and it looses its server connection on SIGSTOP, totally breaking the block devices it provides. What would be the right way to prevent this problem? Fix sysvinit? Fix nbd-client? Note that for nbd-root installations nbd-client is run via (a copy of) /usr/share/initramfs-tools/scripts/local-top/nbd and not via /etc/init.d/nbd-client. Other than that I'm very happy with the stability and maturity of Devuan. Running the installer on a diskless system, installing into a network block device (with cryptsetup and LVM on top), then still being able to boot the system without problems was sort of surprising :) cheers, David  https://github.com/NetworkBlockDevice/nbd/issues/51  https://sourceforge.net/p/nbd/mailman/message/27368126/ -- System Information: Distributor ID: Devuan Description: Devuan GNU/Linux 2.0 (ascii) Release: 2.0 Codename: ascii Architecture: x86_64 Kernel: Linux 4.9.0-6-amd64 (SMP w/4 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Init: sysvinit (via /sbin/init) Versions of packages nbd-client depends on: ii debconf [debconf-2.0] 1.5.61 ii libc6 2.24-11+deb9u3 ii libgnutls30 3.5.8-5+deb9u3 nbd-client recommends no packages. nbd-client suggests no packages. -- debconf information: nbd-client/no-auto-config: nbd-client/killall_set: