[
https://issues.apache.org/jira/browse/HTRACE-118?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14344023#comment-14344023
]
Colin Patrick McCabe commented on HTRACE-118:
---------------------------------------------
I think this makes sense in an HDFS context, since we can have the thread
emptying the buffer in DFSOutputStream set its span's parents to any trace
spans that caused the buffer emptying to happen. I'm going to come up with a
patch for this, but we'll have to wait until htrace 3.2 is released and hadoop
is upgraded to it to commit it, of course.
committed to 3.2
> Java API: support setting the parents of a span after the span is created
> -------------------------------------------------------------------------
>
> Key: HTRACE-118
> URL: https://issues.apache.org/jira/browse/HTRACE-118
> Project: HTrace
> Issue Type: Bug
> Affects Versions: 3.2.0
> Reporter: Colin Patrick McCabe
> Assignee: Colin Patrick McCabe
> Attachments: HTRACE-118.001.patch
>
>
> The Java API should support setting the parents of a span after the span is
> created. This is necessary since some spans may get more parents after they
> start. For example, if we have a write-ahead log that we start flushing when
> a certain number of log entries has accumulated, we may also flush log
> entries that are created after the flush begins.
> To support this case, the Java API should support setting the parents on a
> span. As long as the parents are set before the span is closed, the new
> parents will show up in the JSON we emit.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
