Jean McCormack wrote:
> Ethan Quach wrote:
>>
>>
>> Jean McCormack wrote:
>>> Thanks Sundar. I talked with Sue when she was here last week and 
>>> will keep in touch with her.
>>>
>>> A few more changes to my original email. After talking with Dave 
>>> last week, tftp is going to be
>>> an optional dependency.
>>>
>>> Another thought I've come up with around the SMF stop method. 
>>> Basically what we want to happen
>>> during the stop method is the same as what installadm stop does now. 
>>> However once Sue puts back
>>> we'll want installadm stop minus the setting of status=off in the 
>>> data file. So my thought, with help from Sanjay, is
>>> to create installadm shutdown which will do everything that stop 
>>> does minus the setting of status = off.
>>> That way the code only needs to be maintained in one place.
>>>
>>> Here's what I'm thinking.
>>>
>>> shutdown_service {
>>> stop the process and ai_webserver. This is equivalent to what 
>>> remove_service does now.
>>> }
>>>
>>> remove_service{
>>> shutdown_service()
>>> set status=off
>>> }
>>>
>>> So installadm stop would remain the same in that it calls 
>>> remove_service.
>>> installadm shutdown would call shutdown_service.
>>>
>>> What do you think?
>>
>> Is shutdown private? If so, its merely implementation (cleaner)
>> so I'm okay with that.
> I believe it would need to be another command line subcommand. 
> Otherwise I don't
> know how the smf method could call it. I am open to other suggestions. 
> My goal is
> to not have the same code in two places, setup-service.sh and 
> svc-install_server. That's
> a maintenance nightmare.

Agreed with that.  If it must be from the command line, have you
thought about just adding an option to 'stop'?  'stop -t' perhaps?


thanks,
-ethan

>
> Jean
>
>>
>> -ethan
>>
>>>
>>> Jean
>>>
>>>
>>>
>>> Sundar Yamunachari wrote:
>>>> Jean McCormack wrote:
>>>>> Here's what I'm planning on doing. Please speak up soon if this 
>>>>> doesn't look right.
>>>>> This is the result of an office conversation with Ethan and Evan.
>>>>>
>>>>> The name of the service will be svc://system/install/server:default
>>>>>
>>>>> The method and manifest (svc-install_server and server.xml) will 
>>>>> live in usr/src/cmd/installadm.
>>>>> They will be packaged into SUNWinstalladm-tools.
>>>>>
>>>>> The following will be required dependencies of the service:
>>>>> svc:/network/dns/multicast:default
>>>>> svc:/network/tftp/udp6:default
>>>>> svc:/network/http:apache2
>>>>>
>>>>> The following will be an optional dependency of the service:
>>>>> svc:/network/dhcp-server:default
>>>>>
>>>>>
>>>>> The start method will have an infinite timeout (timeout_seconds='0')
>>>>> The functionality of the start method will be to read the 
>>>>> service_data file
>>>>> and for each "on" service to installadm start <svcname>
>>>>>
>>>>> The stop method will have an infinite timeout (timeout_seconds='0')
>>>>> The functionality of the stop method will be to determine which 
>>>>> services
>>>>> are on and stop the ai webserver for those services. Sundar, how do I
>>>>> determine if the service is on? Ethan commented that he would 
>>>>> rather not
>>>>> trust the service_data file for this since there is the 
>>>>> possibility it might not
>>>>> reflect reality.
>>>> Sue is fixing bugs 6128 (installadm reuses /var/ai data and screw 
>>>> up manifest -- This is data corruption and should be fixed) and 
>>>> 6738 (SUNWinstalladm-tools in b108 clobber existing AI data without 
>>>> rebooting -- Similar to above problem). The fix involves creating 
>>>> one text file for each service (instead of one common file 
>>>> service_data for all services) and adding a status field to 
>>>> indicate whether the service is currently running or stopped. So 
>>>> this functionality will be available. I can help you to figure out 
>>>> how to find that information.
>>>>
>>>> Thanks,
>>>> Sundar
>>>>>
>>>>> Associated work:
>>>>> 1) Remove the enabling of the services listed above from installadm
>>>>> 2) Add the status to the service_data file
>>>>> 3) installadm create-service should be modified to enable this smf 
>>>>> service if it's not.
>>>>>
>>>>>
>>>>> Jean
>>>>> _______________________________________________
>>>>> caiman-discuss mailing list
>>>>> caiman-discuss at opensolaris.org
>>>>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>>>>
>>>
>>> _______________________________________________
>>> caiman-discuss mailing list
>>> caiman-discuss at opensolaris.org
>>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>

Reply via email to