www/obhttpd and question marks in sockstat
m.seaman at infracaninophile.co.uk
Tue Mar 29 12:05:19 UTC 2016
On 2016/03/29 12:28, Niklaas Baudet von Gersdorff wrote:
> I am trying to get www/obhttpd running in a jail. `service obhttpd
> start` works and I get it listening on port 80. But if I run `service
> obhttpd stop` I encounter two problems:
> 1. It seems that there's a problem terminating all related processes.
> Stopping obhttpd.
> kill: 8221: No such process
> kill: 8222: No such process
> kill: 8223: No such process
> kill: 8224: No such process
Looks like the obhttpd process is trying to kill of its children
individually, but they've all already been killed. Possibly by a signal
to the whole process group. It's untidy, but it has achieved your aim
of killing the obhttpd processes.
Try watching what happens in top(1), and tailing the logfile(s) as you
shutdown the obhttpd service. You should see obhttpd and its child
processes disappear as you type 'service obhttpd stop', and the child
process pids should correspond to what you see in the error message.
> 2. I end up with some ports open that I don't know how to close.
> USER COMMAND PID FD PROTO LOCAL ADDRESS FOREIGN ADDRESS
> ? ? ? ? tcp6 2a00:c98:2200:af07:6:0:1:1:80 *:*
> ? ? ? ? tcp4 10.15.1.1:80 *:*
> Some ideas on what could be causing the first and second issue? How can I close
> these ports manually? Is this worth a bug report or a mistake on my side?
This connection is already closed. You can tell by the question marks.
It only still exists in the sockstat output as the kernel is hanging
onto the connection details so it can reap any stray packets that may,
belatedly, arrive. If you wait patiently, these lines will eventually
disappear from the sockstat output. Or you can just restart the daemon
anyhow -- it should come up normally.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 972 bytes
Desc: OpenPGP digital signature
More information about the freebsd-questions