Hide Forgot
Created attachment 556121 [details] dmesg output using systemd-shutdown debug script Description of problem: System takes a long time to shutdown with systemd-37-7.fc16.i686 installed. Version-Release number of selected component (if applicable): systemd-37-7.fc16.i686 How reproducible: Always Steps to Reproduce: 1. Shutdown the system Actual results: Slow shutdown (couple of minutes) Expected results: Shutdown should be relatively fast Additional info: This doesn't happen when I downgrade to systemd-37-6.fc16.i686 Attached is the dmesg output using systemd-shutdown debug script.
How long does "systemctl stop oracle-xe.service" usually take? Does the system shutdown quickly if you stop oracle-xe beforehand? Could you capture the shutdown log once more, but this time after having booted with "log_buf_len=1M systemd.log_level=debug systemd.log_target=kmsg"? Also getting the same log from the working systemd version may help us spot the difference.
So I noticed that it wasn't a downgrade to 37-6 but to what's current in stable, 37-3, where the problem doesn't occur. To answer your questions: 1) [root@jwatzke ~]# time systemctl stop oracle-xe.service real 0m7.367s user 0m0.002s sys 0m0.003s 2) If I shutdown oracle-xe before, it does shutdown faster. 3) I'm attaching debug-systemd-37-8.txt (slightly newer version but still exhibits issue) which has the shutdown log with the debug kernel params turned on. 4) I'm attaching debug-systemd-37-3.txt which has the shutdown log with the debug kernel params turned on for systemd-37-3. I can see in that log systemd is able to correctly shutdown oracle-xe quickly.
Created attachment 556444 [details] systemd-37-8 debug log with debug kernel params on
Created attachment 556445 [details] systemd-37-3 debug log with debug kernel params on
I remember looking at the logs, but I did not figure out anything. Are you still seeing the bug? It could help to setup a debug shell and poke around a bit while the service is stopping. To see what processes are running, perhaps get some backtraces... http://fedoraproject.org/wiki/Systemd_early_debug_shell
Yes this still happens in F17 so I've updated the bug to reflect that. Let me see if I can setup the debug shell and get some more info.
This bug report has been in needinfo status way to long thus closing. Feel free to reopen and providing the requested information. Thanks