Quoth Bernard Biessy on Mon, Dec 11, 2006 at 02:27:27AM -0800:
> I would like to use rsyncd manifest but I have some errors with the methods :
>
> When I want to enable the service, rsync is in maintenance state :
> :svcadm -v enable svc:/network/rsync:
> svc:/network/rsync:default enabled.
> :svcs -a | grep rsync
> maintenance Dec_08 svc:/network/rsync:default
> :
>
> When I check with svcs, I see :
> :svcs -xv
> svc:/network/rsync:default (RSYNC daemon)
> State: maintenance since Fri Dec 08 15:40:44 2006
> Reason: Method failed.
> See: http://sun.com/msg/SMF-8000-8Q
> See: man -M /usr/share/man -s 7 rsync
> See: /var/svc/log/network-rsync:default.log
> Impact: This service is not running.
svcs -x can distinguish failure of the start method, so since the reason
isn't "Start method failed...", I suspect one of the other methods.
> <!-- METHODS -->
> <!-- start & stop methods are required, restart method is optional -->
>
> <exec_method type="method" name="start" exec="/usr/local/bin/rsync
> --daemon
> --config=/usr/local/etc/rsyncd.conf
> --port=873" timeout_seconds="60"/>
>
> <exec_method type="method" name="stop" exec=":kill"
> timeout_seconds="60"/>
>
> <exec_method type="method" name="refresh" exec=":kill -HUP"
> timeout_seconds="60"/>
These look ok.
> <!-- SERVICE MODEL -->
> <!-- It is a transient service - These are often configuration
> services which require no long-running processes in order
> to provide service. -->
> <property_group name="startd" type="framework">
> <propval name="duration" type="astring" value="transient"/>
> </property_group>
Uh oh, why are you declaring the service to be transient? Your start
method launches rsync with --daemon, so I assume it will spawn
a long-running process.
This would cause your stop method -- :kill -- to fail, which explains
the "Reason: Method failed." message. :kill is illegal for transient
services, because no processes are kept track of to kill.
Quoth Bernard Biessy on Mon, Dec 11, 2006 at 06:37:17AM -0800:
> I do :
> :svcadm clear svc:/network/rsync:default
> :svcadm restart svc:/network/rsync:default
> :svcadm enable svc:/network/rsync:default
> :svcs -a | grep rsync
> online 15:21:41 svc:/network/rsync:default
> :
> :svcadm disable svc:/network/rsync:default
> :svcs -a | grep rsync
> maintenance 15:22:23 svc:/network/rsync:default
> :
This confirms it. There wasn't a problem bringing the service online --
the problem occurred when you disabled it. Since you declared the
service to be transient, you should be able to find the process with ps
or pgrep, but it svcs -p won't associate it with the service.
> And in the log file, there is now :
>
> :tail -f /var/svc/log/network-rsync:default.log
> ...
> [ Dec 11 15:20:55 Leaving maintenance because clear requested. ]
> [ Dec 11 15:20:55 Disabled. ]
> [ Dec 11 15:21:41 Enabled. ]
> [ Dec 11 15:21:41 Executing start method ("/usr/local/bin/rsync --daemon
>
> --config=/usr/local/etc/rsyncd.conf --port=873") ]
> [ Dec 11 15:21:41 Method "start" exited with status 0 ]
> [ Dec 11 15:22:23 Stopping because service disabled. ]
> :
Hmm, no useful error messages. Indeed,
http://src.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/cmd/svc/startd/method.c#709
shows that we log an error for this situation, but not to the service's
log file. Check the console, /var/adm/messages, or
/var/svc/log/svc.startd.log for ":kill with no contract". I'll file
a bug to make svc.startd send an error message to the service's log
file, too.
David
