[ 
https://issues.apache.org/jira/browse/UIMA-4883?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15926811#comment-15926811
 ] 

Lou DeGenaro commented on UIMA-4883:
------------------------------------

ducc_services CLI/API until now communicated directly with Services Manager 
(SM) via HTTP.  To both facilitate "single point of entry" and system events 
logging, the CLI/API for services interactions will be made via Orchestrator 
(OR).  

Communications path with be: CLI/API via HTTP to/from OR, then OR via AMQ 
to/from SM.

Changes to realize this outcome:

1. Add new JMS endpoint+type to default.ducc.properties

ducc.sm.api.endpoint=ducc.sm.api
ducc.sm.api.endpoint.type=queue

2. Remove unused HTTP for Service Manager

ducc.sm.http.port=19989

3. Update CLI/API to use HTTP via Orchestrator daemon (OR) for Services daemon  
(SM) requests

4. Update CommonConfiguraton to provide ducc.sm.api.endpoint

5. Update Orchestrator daemon to:
     - field HTTP request from CLI/API (listen for ServiceRequestEvent)
     - forward to SM via AMQ queue (via SmChannel class)
     - receive reply via AMQ
     - send HTTP response back to CLI/API

6. Update Orchestrator daemon to record Services request & results to 
system-events.log

7. Update Services Manager daemon to:
     - stop using HTTP port
     - process requests via AMQ (from Orchestrator)

8. Update Transport to:
     - marshall/unmarshall Events (not HTTP strings)
     - use common ServicesRequestEvent base class for all service requests for 
OR listening convenience


> DUCC system events log
> ----------------------
>
>                 Key: UIMA-4883
>                 URL: https://issues.apache.org/jira/browse/UIMA-4883
>             Project: UIMA
>          Issue Type: Improvement
>          Components: DUCC
>            Reporter: Lou DeGenaro
>            Assignee: Lou DeGenaro
>             Fix For: future-DUCC
>
>
> DUCC should keep a system-wide events log to record for example:
> 1. OR/SM/RM requests/replys
> 2. Job/AP/Service start/end
> 3. Node up/down determination
> 4. Monitor cancellations



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Reply via email to