accumulo-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Josh Elser <josh.el...@gmail.com>
Subject Re: Trace log
Date Fri, 16 May 2014 00:33:34 GMT
Just to be clear, if the Warning message eventually goes away (should be 
within seconds, maybe minutes), then it's probably just asynch delay.

If a substantial time later you're still getting the warning, that's 
probably a sign that the tracing is used wrong (opened and not closed as 
Eric said).

On 5/7/14, 4:55 PM, Eric Newton wrote:
> Your request can cause some spans to start asynchronously.  Also, spans
> are sent to the tracer asynchronously.  The result is that some of the
> spans may not be recorded by the time that your request finishes.  When
> this happens, children of those requests will exist, but their parent
> spans are not found.  The shell retries a few times with a small wait,
> but it will give up and just print the child spans not knowing what
> started them.
>
> It can also be a bug in the trace logic.  A span that was opened, but
> never closed, for example.
>
> -Eric
>
>
>
> On Mon, May 5, 2014 at 7:28 PM, Amit Kapoor <amit@koanect.com
> <mailto:amit@koanect.com>> wrote:
>
>     Hi folks,
>
>        We are trying to use a variation of IntersectingIterator and
>     turned on trace to help us debug
>        certain performance issues (more on that in a separate email).
>     The trace log page is showing
>        this:
>
>             Warning: the following spans are not rooted!
>        49561cfdf8e1ca5e fa33ce2827eb1141 client:startMultiScan
>        c70795aa7b14de02 f6457a7e4060dec4 client:startMultiScan
>        8ecd2bd6e0342911 751d55532acecba7 client:startMultiScan
>        74a8d66d235750b2 b82977a699403c46 client:startMultiScan
>        ...
>
>        Any hints on what is going on here?
>
>        regards
>
>     Amit
>
>

Mime
View raw message