[
https://issues.apache.org/jira/browse/HADOOP-6884?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12900390#action_12900390
]
Doug Cutting commented on HADOOP-6884:
--------------------------------------
> I found at least two three-variable cases in Erik's latest patch.
I missed those, but it's still different from your benchmark, which focussed on
boxed variables. Those examples might look something like:
debug("Exception while invoking {} of {}. Retrying.", method.getName(),
implementation.getClass(), StringUtils.stringifyException(e));
That last parameter would make things slow, but, I'd be surprised if you can
show that optimizing this substantially improves Hadoop performance, since this
only runs in the case that a retryable method fails with an exception. If we
ever found that such cases were problematic, we might add methods like:
static void debug(String message, Throwable t);
static void debug(String message, Throwable t, Object p1);
static void debug(String message, Throwable t, Object p1, Object p2);
static void debug(String message, Throwable t, Object... parameters);
Perhaps we could submit a patch to slf4j adding such methods there.
The second case could become something like:
debug("for protocol authorization compare ({}):{} with ", clientPrincipal,
shortName, user.getShortUserName));
In this case all parameters are strings, so no boxing would occur.
> Your benchmark also shows that LOG.isDebugEnabled() provides the best results.
It provides significantly better results in a micro-benchmark. But
optimizations that increase code size should show significant improvement in
real application performance, not just micro-benchmarks. Until someone's shown
that this optimization measurably improves end performance, I see no urgency to
commit it.
> Add LOG.isDebugEnabled() guard for each LOG.debug("...")
> --------------------------------------------------------
>
> Key: HADOOP-6884
> URL: https://issues.apache.org/jira/browse/HADOOP-6884
> Project: Hadoop Common
> Issue Type: Improvement
> Affects Versions: 0.22.0
> Reporter: Erik Steffl
> Assignee: Erik Steffl
> Fix For: 0.22.0
>
> Attachments: FunAgain.java, FunAgain.java, HADOOP-6884-0.22-1.patch,
> HADOOP-6884-0.22.patch
>
>
> Each LOG.debug("...") should be executed only if LOG.isDebugEnabled() is
> true, in some cases it's expensive to construct the string that is being
> printed to log. It's much easier to always use LOG.isDebugEnabled() because
> it's easier to check (rather than in each case reason whether it's necessary
> or not).
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.