can't kill dump process
le...@yahoo.com
my filesystems. For some reason there is a stuck dump
process on my BSD/OS 3.1 system that I can't kill! The
tape has been ejected, but "kill -9 16120" does no good.
I want to avoid rebooting this machine, so can anyone help?
# ps auxw | grep dump
root 16120 0.0 0.2 584 300 ?? DW 23Apr99 0:00.02 dump 0uBbf 20000000 32 /dev/nrst0
/
_____________________________________________________________
Do You Yahoo!?
Free instant messaging and more at http://messenger.yahoo.com
--
To unsubscribe, send 'unsubscribe' to bsdi-user...@mailinglists.org
Archives of this list can be found at: http://www.nexial.com/mailinglists/
Alan Clegg
> my filesystems. For some reason there is a stuck dump
> process on my BSD/OS 3.1 system that I can't kill! The
> tape has been ejected, but "kill -9 16120" does no good.
> I want to avoid rebooting this machine, so can anyone help?
>
> # ps auxw | grep dump
> root 16120 0.0 0.2 584 300 ?? DW 23Apr99 0:00.02 dump 0uBbf 20000000 32 /dev/nrst0
> /
Processes waiting for device IO to complete won't die until the
pending IO finishes. Since the tape is out of the drive, it'll never
happen.
Reboot.
AlanC
System Administrator
Try doing a "kill -TERM 16120" or a "kill -KILL 16120"
Andrew Hall
On Wed, 19 May 1999 le...@yahoo.com wrote:
> I have "dump" running from cron once per week to backup
> my filesystems. For some reason there is a stuck dump
> process on my BSD/OS 3.1 system that I can't kill! The
> tape has been ejected, but "kill -9 16120" does no good.
> I want to avoid rebooting this machine, so can anyone help?
>
> # ps auxw | grep dump
> root 16120 0.0 0.2 584 300 ?? DW 23Apr99 0:00.02 dump 0uBbf 20000000 32 /dev/nrst0
> /
>
> _____________________________________________________________
> Do You Yahoo!?
> Free instant messaging and more at http://messenger.yahoo.com
Nathan Price
>> I have "dump" running from cron once per week to backup
>> my filesystems. For some reason there is a stuck dump
>> process on my BSD/OS 3.1 system that I can't kill! The
>> tape has been ejected, but "kill -9 16120" does no good.
>> I want to avoid rebooting this machine, so can anyone help?
>>
>> # ps auxw | grep dump
>> root 16120 0.0 0.2 584 300 ?? DW 23Apr99 0:00.02 dump
0uBbf 20000000 32 /dev/nrst0
>> /
>
>pending IO finishes. Since the tape is out of the drive, it'll never
>happen.
>
>Reboot.
Is there no trick for tape devices like that for serial ports? I will
occasionally get processes doing serial port I/O that get stuck waiting for
I/O or flow control and then cannot be killed until I issue a "stty -f
/dev/ttyxx flushout". This clears the pending I/O and releases the serial
device, allowing the process to be killed and restarted.
Nathan
Henry Spencer
>pending IO finishes. Since the tape is out of the drive, it'll never
>happen.
>
>Reboot.
And report it as a bug. Device drivers should not wait forever for events
that are supposed to take a finite amount of time. (Granted, for tape
devices, timeouts have to be set generously, because tape rewinds and such
*can* be lengthy... but the driver nevertheless should *eventually* decide
that the completion interrupt must have gone astray somehow, and cope.)
> Is there no trick for tape devices like that for serial ports?
On half-inch tape drives, typically you could trigger an interrupt by
using the drive's manual controls to wind the tape ahead somewhat while
offline, and then starting a rewind and putting the drive online before
the rewind finished. This is a bit more difficult, to put it mildly, on
modern tape drives whose only control is an EJECT button. You can try
inserting and then ejecting a scratch tape, but it's unlikely to help.
Henry Spencer
he...@spsystems.net
(he...@zoo.toronto.edu)