Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Flume >> mail # dev >> Review Request: FLUME-1842.AsyncHBaseSink timeout is not calculated correctly


Copy link to this message
-
Re: Review Request: FLUME-1842.AsyncHBaseSink timeout is not calculated correctly

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/8968/#review15412
-----------------------------------------------------------
Hari, the patch looks good! A few comments below.
flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/AsyncHBaseSink.java
<https://reviews.apache.org/r/8968/#comment33207>

    nit: Is there a better variable name? Perhaps sinkCallbackPool?

flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/AsyncHBaseSink.java
<https://reviews.apache.org/r/8968/#comment33199>

    nit: variable should be named isDelayTesting or isTimeoutTesting

flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/AsyncHBaseSink.java
<https://reviews.apache.org/r/8968/#comment33201>

    Just curious as to why this was added or perhaps why it was missing? I am not too familiar with the async hbase api.

flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/AsyncHBaseSink.java
<https://reviews.apache.org/r/8968/#comment33203>

    Curious why we converted to nano seconds from milleseconds? IMHO ms are easier to comprehend

flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/AsyncHBaseSink.java
<https://reviews.apache.org/r/8968/#comment33214>

    It looks like it's possible for timeRemaining to be negative. That is if the condition returned with 1 nanosecond left and then it took longer than to to check the callbacks. Condition.await() java doc doesn't say what occurs with a negative number.

flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/AsyncHBaseSink.java
<https://reviews.apache.org/r/8968/#comment33196>

    How come we are converting back and forth between nanos and millis?

flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/AsyncHBaseSink.java
<https://reviews.apache.org/r/8968/#comment33195>

    nit: Should catch InterruptedException only

flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/AsyncHBaseSink.java
<https://reviews.apache.org/r/8968/#comment33193>

    Same as above

flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/AsyncHBaseSink.java
<https://reviews.apache.org/r/8968/#comment33194>

    nit: Should catch InterruptedException only
- Brock Noland
On Jan. 16, 2013, 8:47 p.m., Hari Shreedharan wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/8968/
> -----------------------------------------------------------
>
> (Updated Jan. 16, 2013, 8:47 p.m.)
>
>
> Review request for Flume.
>
>
> Description
> -------
>
> Made changes to calculate remaining time and wait only for that long
>
>
> This addresses bug FLUME-1842.
>     https://issues.apache.org/jira/browse/FLUME-1842
>
>
> Diffs
> -----
>
>   flume-ng-doc/sphinx/FlumeUserGuide.rst 58a115e
>   flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/AsyncHBaseSink.java 6b34873
>   flume-ng-sinks/flume-ng-hbase-sink/src/main/java/org/apache/flume/sink/hbase/HBaseSinkConfigurationConstants.java fad026c
>   flume-ng-sinks/flume-ng-hbase-sink/src/test/java/org/apache/flume/sink/hbase/TestAsyncHBaseSink.java 1f61406
>
> Diff: https://reviews.apache.org/r/8968/diff/
>
>
> Testing
> -------
>
> Added unit test. Current tests pass
>
>
> Thanks,
>
> Hari Shreedharan
>
>