Bug #1047
closed
rcK shutdown gets hung
100%
Description
Running reboot will call /etc/init.d/rcK because of the shutdown config in inittab.
But rcK gets hung on something. Add debug code to it and run it manually to see what's hanging it.
Updated by Hammel 6 months ago
- Status changed from New to In Progress
- % Done changed from 0 to 10
First problem: rcK runs the list in the wrong order. It should run them in reverse numeric order. This can be fixed by listing the files and then reverse sorting with "sort -r".
Second problem: it gets stuck shutting down the network.
$ /etc/init.d/rcK Splash framebuffer: 0 /etc/init.d/S01logging: line 14: Stopping logging: : not found /etc/init.d/S01logging: line 17: OK: not found Stopping syslogd: OK Stopping klogd: FAIL 1+0 records in 1+0 records out 32 bytes (32B) copied, 0.000244 seconds, 128.1KB/s Stopping system message bus: killall: dbus-daemon: no process killed done Stopping rpcbind daemon: OK Stopping cron. stopped process in pidfile '/var/run/crond.pid' (pid 725) Stopping iptables: OK Stopping network...
This eventually kicked me out, but the system was still running - I was in musicfe and was able to hit keys and have it respond. If I then exited musicfe, the launcher displayed but eventually keyboard input was lost. Eventually keyboard input came back and I could move around the launcher interface. When I tried to use Ctrl-T to get a terminal, it locked up again. Eventually keyboard came back again. This time I tried to select terminal app. Same lockup problem.
I'm going to implement the reverse order shutdown first, then try and see where it stops.
Updated by Hammel 6 months ago
Now it looks like this from the ssh console:
$ /etc/init.d/rcK Shutting down SMB services: OK Shutting down NMB services: OK Stopping PiBox Daemon...OK Stopping Bluetooth daemon bluetoothd... killall: bluetoothd: no process killed Shutting down NFS mountd: OK Shutting down NFS daemon: OK Shutting down NFS services: OK Stopping NFS statd: OK Stopping ALSA... Stopping lighttpd: OK Stopping dropbear sshd: OK Stopping ntpd: OK Stopping network...
But since ssh was killed, that doesn't tell me what I need to know. So I need to run this from the target system directly, redirecting output to persistent storage.
Updated by Hammel 6 months ago
- % Done changed from 10 to 80
Well, now it looks fine.
$ cat rck.log Shutting down SMB services: OK Shutting down NMB services: OK Stopping PiBox Daemon...OK Stopping Bluetooth daemon bluetoothd... Shutting down NFS mountd: OK Shutting down NFS daemon: OK Shutting down NFS services: OK Stopping NFS statd: OK Stopping ALSA... Stopping lighttpd: OK Stopping dropbear sshd: OK Stopping ntpd: OK Stopping network... Stopping iptables: OK Stopping cron. stopped process in pidfile '/var/run/crond.pid' (pid 696) Stopping rpcbind daemon: OK Stopping system message bus: done Stopping klogd: OK Stopping syslogd: OK Splash framebuffer: 0
Now when I run "reboot" manually it reboots nearly immediately. Just reversing the order appears to be the only fix required.
I will note that on the test target, a reboot happens quickly but the network is not bringing up eth0 on the next startup. This might be just this unit, however. Will test on other hardware.