Re: [Help-gnu-radius] Hung children

[John Stone]

Sergey,

S UID        PID  PPID  C PRI  NI  RSS    SZ WCHAN  STIME TTY
           TIME CMD
S root     17291     1  0  69   0 1236   648 do_sel 14:18 ?
       00:00:00 /usr/local/radius/sbin/radiusd
R root     17618 17291 47  80   0 1444   649 -      14:25 ?
       00:03:26 /usr/local/radius/sbin/radiusd
R root     17712 17291 32  79   0 1444   649 -      14:27 ?
       00:01:48 /usr/local/radius/sbin/radiusd
R root     17766 17291 27  80   0 1444   649 -      14:28 ?
       00:01:12 /usr/local/radius/sbin/radiusd
R root     17878 17291 24  80   0 1444   649 -      14:30 ?
       00:00:45 /usr/local/radius/sbin/radiusd
--- from log file ---
Mar 01 14:28:57: Main.notice: Dropping AUTH packet from client 207.100.33.8, ID: 187: duplicate request
Mar 01 14:29:57: Main.notice: Killing unresponsive AUTH child pid 17766
Mar 01 14:29:57: Auth.crit: CHILD: exit on signal (11)

Process 17766 (and the others) is still there running up processor load. Disk activity
appears to be OK. This box serves a number of other daemons which all seem to
be working as expected.

Thanks

Sergey Poznyakoff wrote:

> I see messages in radius.log about:
>     "Killing unresponsive AUTH child pid nnnn"
> but the process(s) never goes away.

What do you see in ps output then? Killing hung child may fail
only in one case: when this child entered uninterruptible sleep.
In this case, you will see D in the STAT column of ps output.
Usually this means some problems with IO devices (e.g. hard
disks).

Regards,
Sergey

_______________________________________________
Help-gnu-radius mailing list
address@hidden
http://mail.gnu.org/mailman/listinfo/help-gnu-radius

--
John Stone