That is practically what I meant. Maybe with an exception to the 
"returns right away" part. We have somehow similar "spawn" flag used for 
tc:start and perhaps it should be used for cargo containers too, so user 
could see the log on the console and Ctrl-C containers when needed.

  I wasn't sure if we should have "cargo" word in our goal names, but 
referencing cargo in tc:cargo-start vs. tc:container-start can be 
actually good thing. So, it is still open for discussion. :-)

  regards,
  Eugene


Hung Huynh wrote:
> Or we can something like this?
>
> tc:cargo-start --> returns right away after appservers start
> tc:cargo-stop --> stop all appservers
>
> Hung-
>
> Eugene Kuleshov wrote:
>> Steven Harris wrote:
>>   
>>> Can we stop it just by keeping track of the processes and then  
>>> killing them? 
>>>     
>>   Killing them is not a problem. When to kill - is. :-)
>>   For instance, when using tc:run I currently assume that all started 
>> processes will eventually complete and exit. That won't be the case for 
>> the web application servers.
>>   
>>> I'm assuming cargo doesn't have
>>> some sort of standard shutdown thing that each container that it  
>>> controls can implement.
>>>   
>>>     
>>   Actually it does. That is why I am going to add tc:container-stop 
>> goal. See example in my first email.
>>
>>   Another useful improvement could be to optionally pass list of node 
>> names to start/stop with tc:run or tc:container-start/stop, so we could 
>> have configuration in a single pom, but split actual processes across  
>> multiple physical nodes. For example:
>>
>>   mvn -DnodeNames=master tc:run
>>   mvn -DnodeNames=worker1,worker2 tc:run
>>   mvn -DnodeNames=worker3,worker4 tc:run
>>
>>   regards,
>>   Eugene
>>
>>     
>

_______________________________________________
tc-dev mailing list
[email protected]
http://lists.terracotta.org/mailman/listinfo/tc-dev

Reply via email to