hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Appy (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (HBASE-16864) Procedure v2 - Fix StateMachineProcedure support for child procs at last step
Date Tue, 18 Oct 2016 20:20:00 GMT

    [ https://issues.apache.org/jira/browse/HBASE-16864?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15586557#comment-15586557
] 

Appy edited comment on HBASE-16864 at 10/18/16 8:19 PM:
--------------------------------------------------------

Now that there is a state to mark end of flow, it seems that Flow itself is redundant. Procedures
can just set their next state and don't have to explicitly say more states/no more states.
{noformat}
switch(state){
  case state1:
    do stuff;
    setNextState(state2);
  case state2:
    do stuff;
    setNextState(EOF_STATE);
}
{noformat}
Can be done separately. But does it sound right?



was (Author: appy):
Now that there is a state to mark end of flow, it seems that Flow itself is redundant. Procedures
can just set their next state and don't have to explicitly say more states/no more states.

switch(state){
  case state1:
    do stuff;
    setNextState(state2);
  case state2:
    do stuff;
    setNextState(EOF_STATE);
}

Can be done separately. But does it sound right?


> Procedure v2 - Fix StateMachineProcedure support for child procs at last step
> -----------------------------------------------------------------------------
>
>                 Key: HBASE-16864
>                 URL: https://issues.apache.org/jira/browse/HBASE-16864
>             Project: HBase
>          Issue Type: Sub-task
>          Components: proc-v2
>    Affects Versions: 2.0.0
>            Reporter: Matteo Bertozzi
>            Assignee: Matteo Bertozzi
>             Fix For: 2.0.0
>
>         Attachments: HBASE-16864-v0.patch
>
>
> There is a bug in the StateMachineProcedure when we add child procs in the last step.
On recovery we end up spinning on the last step without ever completing. the fix is to introduce
an eof step so recovery knows that we are already done once the all the children are terminated.




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message