[
https://issues.apache.org/jira/browse/HTRACE-5?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14247327#comment-14247327
]
Colin Patrick McCabe commented on HTRACE-5:
-------------------------------------------
Hi [~billie.rinaldi],
Maybe I'm missing something, but this patch doesn't seem correct to me. The
code in {{TraceRunnable.java}} looks correct as-is. It creates the trace scope
before executing the {{Runnable}}, and closes the trace scope once the runnable
is done (in the {{finally}} block.) It's not up to {{TraceRunnable}} to
determine if there are any other trace spans that the current thread knows
about, or to close trace spans other than the current one that it's managing.
Just guessing here, but it sounds like you have other trace spans, not created
by the {{TraceRunnable}} code, which you may have to also close. Can you check
for this in your code?
> Tracing never ends when using TraceRunnable in a thread pool
> ------------------------------------------------------------
>
> Key: HTRACE-5
> URL: https://issues.apache.org/jira/browse/HTRACE-5
> Project: HTrace
> Issue Type: Bug
> Reporter: Billie Rinaldi
> Attachments: HTRACE-5.1.patch
>
>
> Accumulo recently started using htrace, and when we tried to use the
> TraceExecutorService / TraceRunnable / TraceCallable we noticed that traces
> seemed to continue indefinitely. This was resolved when we added
> Tracer.getInstance().continueSpan(null) after stopping the span. See
> https://github.com/apache/accumulo/tree/master/core/src/main/java/org/apache/accumulo/core/trace/wrappers
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
