[
https://issues.apache.org/jira/browse/HTRACE-69?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14679479#comment-14679479
]
Masatake Iwasaki commented on HTRACE-69:
----------------------------------------
Hi [~leftnoteasy]. Thanks for your input!
{noformat}
- Why my AM takes so much time to be allocated/launched
- Why containers take so much time to be allocated/launched
{noformat}
Current strategy for tracing is collecting all downstream tracing spans at
runtime and searching/filtering at viewing them later on viewer. I think it
would work for these use cases. We can start tracing spans for each (or some
percentage of) container allocation/launching and search slow cases by its
duration. Tracing could be started by configuration. DFSClient already have
similar built-in per-operation tracing.
Actually the reason I filed this jira is the fact that I saw many small RPC
calls on testing some tracing in YARN-2467. It would be useful some sampling on
runtime if you trace whole YARN application having relatively long life span.
The next release of HTrace (htrace-4.0.0) will make it easier to trace specific
modules/products only.
> Filtering child spans by sampler
> --------------------------------
>
> Key: HTRACE-69
> URL: https://issues.apache.org/jira/browse/HTRACE-69
> Project: HTrace
> Issue Type: New Feature
> Reporter: Masatake Iwasaki
> Assignee: Masatake Iwasaki
>
> Trace#startSpan respect the sampler given as argument only when there is no
> ongoing span (i.e. when creating new root span).
> {code}
> public static TraceScope startSpan(String description, Sampler<TraceInfo>
> s, TraceInfo tinfo) {
> Span span = null;
> if (isTracing() || s.next(tinfo)) {
> {code}
> Adding API starting span if {{(isTracing() && s.next(tinfo))}} enables
> filtering of child spans.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
