Ethan Quach wrote:
> Jean,
>
> svc-install-server - these changes don't completely capture
> what I commented on.  I looked, and we do deliver a skeleton
> AI_HTTPD_CONF file, so checking for its existence, though
> necessary  for the invocation of apache to be sane, isn't enough.
> We need to maybe set a flag in the loop that processes install
> services, after line 59, to indicate that we've indeed started at
> least one install service.  Only then should we run our apache
> process, right?  Or is there any other reason to start the
> apache process when they system isn't serving any install
> services?
My thinking was that this was similar to tftp running and dns running
in that it's a process we will need and starting it up is  part of 
initializing
the system for us. i.e. the things we need will be up and running.

Jean
>
>
> -ethan
>
>
>
> Jean McCormack wrote:
>> Webrev is now updated. Sorry it took so long. I wanted to get 
>> everyone's requested changes in, have dinner,
>> and do some testing.
>>
>> Here's the link:
>>
>> http://cr.opensolaris.org/~jeanm/slim_4488_2/
>>
>> Jean
>>
>> Joseph J. VLcek wrote:
>>> Jean McCormack wrote:
>>>> Joseph J VLcek wrote:
>>>>> Jean McCormack wrote:
>>>>>>
>>>>>> Can I please get a code review for 4488
>>>>>>
>>>>>> CR:
>>>>>> http://defect.opensolaris.org/bz/show_bug.cgi?id=4488
>>>>>>
>>>>>> Webrev:
>>>>>> http://cr.opensolaris.org/~jeanm/slim_4488/
>>>>>>
>>>>>> Jean
>>>>>> _______________________________________________
>>>>>> caiman-discuss mailing list
>>>>>> caiman-discuss at opensolaris.org
>>>>>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>>>>>
>>>>>
>>>>> usr/src/cmd/installadm/installadm.c
>>>>> -----------------------------------
>>>>>
>>>>> nit 1:
>>>>> ------
>>>>>
>>>>> 783  * do_enable will restart the specified service
>>>>>
>>>>> Should that read "enbale" instead of "restart"?
>>>>>
>>>>> e.g.:
>>>>>
>>>>> * do_enable will enable the specified service
>>>> Done.
>>>>
>>>>>
>>>>> nit 2:
>>>>> ------
>>>>>
>>>>> @ line 850 add a comment to do_disable() to describe the action 
>>>>> that specifying the -t will have.
>>>>>
>>>> Done.
>>>>>
>>>>> usr/src/pkgdefs/SUNWinstalladm-tools/prototype_com
>>>>> --------------------------------------------------
>>>>>
>>>>> nit:
>>>>> blank lines @ 113 & 120 not needed
>>>> Done
>>>>>
>>>>> issue:
>>>>> Don't need line 114. It's done at line 99
>>>>> 114 d none var 755 root sys
>>>>>
>>>> Gone.
>>>>> usr/src/cmd/installadm/server.xml
>>>>> ----------------------------------
>>>>>
>>>>> Question:
>>>>> ---------
>>>>>
>>>>> Do you want this enabled at install?
>>>>>
>>>>> 40         <create_default_instance enabled='false' />
>>>> I think not.
>>>>>
>>>>> Question:
>>>>> ---------
>>>>>
>>>>> Why udp6? I don't know, is there udp4 tftp?
>>>>>
>>>>> 52                 name='udp6'
>>>> OpenSolaris appears to ship udp6 and not 4.
>>>>>
>>>>> Question:
>>>>> ---------
>>>>>
>>>>> Do you want 0 for timeouts? Doesn't this mean it will wait forever?
>>>>>
>>>>> 63                 timeout_seconds='0' />
>>>>> 69                 timeout_seconds='0' />
>>>> Yes. The issue is that we loop through potentially hundreds of 
>>>> services on
>>>> start and stop. That means doing an installadm enable of each 
>>>> service on
>>>> start and a installadm disable -t on stop. This can take quite a 
>>>> long time.
>>>> There isn't really a value I could pick that would not potentially 
>>>> cause a
>>>> false error.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> usr/src/cmd/installadm/svc-install-server
>>>>> -----------------------------------------
>>>>>
>>>>> Issue 1:
>>>>> --------
>>>>>
>>>>> Style thing.
>>>>>
>>>>> In looking at other files in /lib/svc/method it seems different 
>>>>> conventions are used.
>>>>>
>>>>> But here is ksh93 how to:
>>>>> http://installzone-wiki.central.sun.com/wiki/index.php/Ksh93_Tips
>>>>>
>>>>> I wouldn't hold this up until it conforms but just so you know in 
>>>>> the future.
>>>>>
>>>>> Issue 2:
>>>>> --------
>>>>>
>>>>> If you are going to define the commands in the style:
>>>>>
>>>>> COMMAND=<path>
>>>>>
>>>>> Add echo to @ line 35
>>>> Done
>>>>>
>>>>> e.g.
>>>>> ECHO=/usr/bin/echo
>>>>>
>>>>> Then user ${ECHO} instead of echo.
>>>>>
>>>>>
>>>>> Issue 3:
>>>>> --------
>>>>>
>>>>> Using {} around env variables is safer.
>>>>>
>>>>> You can do it for all if you like... See the Ksh93_Tips for 
>>>>> guidelines.
>>>>>
>>>>> but for example:
>>>>> 81 status=`$GREP "status=" $i | $SED 's/status=//'`
>>>>>
>>>>> To:
>>>>> tatus=`${GREP} "status=" ${i} | ${SED} 's/status=//'`
>>>> Done.
>>>>>
>>>>>
>>>>> Issue 4:
>>>>> --------
>>>>> Maybe you don't need to echo success info. It could fill the log.
>>>>> e.g.:
>>>>> 54                                 echo "Starting $service_name"
>>>>
>>>> Removed those.
>>>>
>>>> Thanks for the review.
>>>>
>>>> Jean
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>>>
>>> Jean,
>>>
>>> Did you update the webrev for me to confirm? It doesn't seem to be 
>>> updated. Am I looking at the wrong URL?
>>>
>>> You should commit before making the webrev so the commit comment is 
>>> in the webrev. always used to forget to do this. ;) I find it does 
>>> help when one is looking at more than one webrev. ;)
>>>
>>> usr/src/cmd/installadm/svc-install-server
>>> -----------------------------------------------------------------
>>> You don't need the PATH env in if you define the COMMAND=<path> env 
>>> vars.
>>>
>>>
>>> Joe
>>
>> _______________________________________________
>> caiman-discuss mailing list
>> caiman-discuss at opensolaris.org
>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss


Reply via email to