[Opendnssec-user] Spec file and changes to ods-control startup script

AndyH at nominet.org.uk AndyH at nominet.org.uk
Thu Nov 26 09:47:44 UTC 2009

> Alexd at nominet.org.uk wrote:
> >> > perhaps it would be better to check if the processes actually died,
> >>
> >> Yes, a much better idea, so that when ods-signer exits the processes
> >> actually gone.  The downside to this is what to do if/when the signer
> >> doesn't exit cleanly
> This is also a problem if you use sleep 1.
> > I've had a few experiences where issuing the "ods-control stop" command
> > stops neither the ods-enforcerd nor the ods-signerd. It would be great
> > if the "ods-control stop" command did always stop both these processes
> > (and checked).
> ods-signerd might not stop if one of the workers is busy (one of the
> signer tools is still running). But the signal is received and the
> signer engine will shutdown eventually (as soon as all workers are
> ready). This might take up to a few minutes if you are signing a large
> zone at the moment.

Sadly the reality is that most sysadmins (in a sample of 1) would just
issue a kill if the process hasn't died within a few seconds.  There is
also the problem of rebooting a server.  A good sysadmin will do an init 6
which will try and stop the signer.  Having to wait minutes is not the
desired option.

Is there any way that the signer can cleanly and quickly abort any tasks
that are underway?


More information about the Opendnssec-user mailing list