Jmockit has worked well for both mocking and stubbing for me. My problem was
System.currentTimeMillis and if you can patch that you can patch most anything.
Sent from my iPhone
On Feb 22, 2012, at 9:23 AM, Ted Yu yuzhih...@gmail.com wrote:
Benoit's comment is directly related to our
Actually jmockit uses byte code patching so you may suffer less reflection
overhead than expected. My guess is that powermock is doing something quite
similar.
Sent from my iPhone
On Feb 22, 2012, at 9:29 AM, Jesse Yates jesse.k.ya...@gmail.com wrote:
Only long standing problem I've found
As long as vendors are being mentioned, it should be pointed out that
MapR's NFS capability also provides this capability.
Followups should be off-list since this is off-topic relative to hbase.
On Sat, Dec 10, 2011 at 11:27 PM, David Engfer david.eng...@gmail.comwrote:
With Cloudera's Hue
Todd,
I am curious what you mean here. How is adding a test suite better than
annotating different tests in the TestNG or Junit 4 style?
On Wed, Nov 2, 2011 at 5:25 PM, Todd Lipcon t...@cloudera.com wrote:
Adding a system test suite would do us some good here. Accumulo has a
very nice one
On Wed, Nov 2, 2011 at 5:33 PM, Todd Lipcon t...@cloudera.com wrote:
On Wed, Nov 2, 2011 at 5:31 PM, Ted Dunning tdunn...@maprtech.com wrote:
Todd,
I am curious what you mean here. How is adding a test suite better than
annotating different tests in the TestNG or Junit 4 style
No. The problem is that you want to emulate real-world behavior which is
probably closer to 1000 threads each doing a single transaction and yielding
than anything else. For instance, if your traffic originates from a
web-farm, each transaction will be in a thread that yields when it finishes
The book is already slightly out of date.
Maven 3 is better and handles almost all maven 2 builds pretty seamlessly.
I don't see a problem with moving forward rather than back.
On Sun, Oct 9, 2011 at 1:03 PM, Jesse Yates jesse.k.ya...@gmail.com wrote:
So the book says to use Maven 2, rather
).
If that's not the case anymore I'll update the book.
On 10/9/11 4:31 PM, Ted Dunning tdunn...@maprtech.com wrote:
The book is already slightly out of date.
Maven 3 is better and handles almost all maven 2 builds pretty seamlessly.
I don't see a problem with moving forward rather than back
update the book.
On 10/9/11 4:31 PM, Ted Dunning tdunn...@maprtech.com wrote:
The book is already slightly out of date.
Maven 3 is better and handles almost all maven 2 builds pretty
seamlessly.
I don't see a problem with moving forward rather than back.
On Sun, Oct 9, 2011 at 1:03
I can get some sponsorship going on my end as well.
On Sun, Oct 2, 2011 at 12:09 AM, Ted Yu yuzhih...@gmail.com wrote:
I agree. We should share the payment.
On Sat, Oct 1, 2011 at 5:05 PM, Todd Lipcon t...@cloudera.com wrote:
Thanks, Andrew! Let us know if we can chip in for the dues.
This sounds like you will have about 500 million rows in your database after
6 months. To my mind, this is at the level of inconvenient for a
conventional database, but hardly impossible.
HBase will definitely hold this much data. It would probably help you to do
some slightly clever tricks to
Hbase offers co-processors which should be able to do this.
And median *can* be accumulated in a small amount of memory. It is a little
trickier than mean, but still doable.
On Tue, Sep 6, 2011 at 11:21 AM, Duane Moore duane.mo...@issinc.com wrote:
- Aggregation
Accumulo offers the ability
There are additional off-shoots of Hadoop that can specifically address
real-time needs such as Spark, S4 and Hstreaming.
Most real-time-ish applications come, however, with a 100% uptime guarantee.
Most simply put, a system that is down and is going to take 10's to 100's
of minutes to come back
Oh lordie, let's not. Logo discussions drown out everything else and never
result in getting a better logo.
At most, how about just having an up or down vote on this one logo. Then
everybody can get on with something important.
On Thu, Sep 1, 2011 at 10:46 PM, Stack st...@duboce.net wrote:
to highly varied evaluations.
Few people have huge variation in problem size and thus can focus on the
interplay between resources and run-time at a (nearly) constant problem
size.
Thanks,
Jon
On Mon, Aug 15, 2011 at 4:53 PM, Ted Dunning tdunn...@maprtech.comwrote:
Sure.
The attached
Actually, I think that anybody who maintains any kind of reference to a JIRA
whether in a distribution or in their own head would like the meaning of a
JIRA to be relatively static.
So even if the community doesn't explicitly know that they care about this,
I bet they care about the consequences.
On Mon, Jul 18, 2011 at 9:32 AM, Jason Rutherglen
jason.rutherg...@gmail.com wrote:
My gut is that this would be a maintenance headache
What specifically do you think would cause a problem?
Tracking versions for one. Everybody has a different favorite. That is the
nice thing about
On Sat, Jul 16, 2011 at 6:28 PM, Jason Rutherglen
jason.rutherg...@gmail.com wrote:
Running the DataNode inside of an HBase process seems like this could
be a good option to enable?
My gut is that this would be a maintenance headache.
Specifically because it would reduce the number of
jason.rutherg...@gmail.com
wrote:
I'm a little confused, I was told none of the HBase code changed with MapR,
if the HBase (not the OS) block cache has a JNI implementation then that
part of the HBase code changed.
On Jul 9, 2011 11:19 AM, Ted Dunning tdunn...@maprtech.com wrote:
MapR does help
MapR does help with the GC because it *does* have a JNI interface into an
external block cache.
Typical configurations with MapR trim HBase down to the minimal viable size
and increase the file system cache correspondingly.
On Fri, Jul 8, 2011 at 7:52 PM, Jason Rutherglen
No.
But you certainly can draw a lot of attention to something you didn't mean
to say.
On Sat, Jun 25, 2011 at 1:02 PM, Todd Lipcon t...@cloudera.com wrote:
Once you say something on the internet, you can't unsay it :)
Yes. New thrift since 0.6 is all byte buffers (thank goodness).
But the hbase dependency is 0.5.0 which is before the great change.
On Tue, May 31, 2011 at 8:03 PM, Stack st...@duboce.net wrote:
Isn't thrift all bytebuffers now? Are you using a new thrift but
your target is 0.89 which may
Stack,
We will hoist a few in your honor. You don't need to feel bad.
On Thu, May 26, 2011 at 2:46 PM, Stack st...@duboce.net wrote:
I find this topic completely inappropriate for dev list. Discussions
of a group of hbasers hanging together in exotic locations, and
reading between the
You should be able to set split points for every customer/date combination
of interest. This would allow you to localize the data to be deleted to
single regions.
On Sat, May 14, 2011 at 12:25 PM, Ophir Cohen oph...@gmail.com wrote:
About the partitionion: I talked about something more
Todd,
I see ycsb on your list.
Where did that go? We have been beating on it as well and have pretty much
decided that it is worthless as it stands.
My thought is that we need a multi-node version that takes directions about
what load to generate via ZK. That is better than a map-reduce based
the map-phase of your program.
On Sat, Mar 26, 2011 at 3:55 PM, Todd Lipcon t...@cloudera.com wrote:
On Sat, Mar 26, 2011 at 3:53 PM, Ted Dunning tdunn...@maprtech.comwrote:
Hmm...
Yeah. I hear that scrapping YCSB meme a lot.
Do you not worry about verifying intermediate results when over
Excuse me?
How does that affect the issue of snapshotting a table?
And how can replication NOT involve meta-data?
On Thu, Mar 17, 2011 at 10:55 PM, jiangwen w wjiang...@gmail.com wrote:
replication does not involve meta data.
On Fri, Mar 18, 2011 at 12:32 PM, Ryan Rawson ryano...@gmail.com
When does ZK plan to adopt this extension?
In general, I agree with Ryan that ZK is a good coordination layer, but the
data (and associated meta-data should be self-hosted to simplify the
consistency problem).
On Thu, Mar 17, 2011 at 9:54 PM, jiangwen w wjiang...@gmail.com wrote:
yes , it is
I already did that. I spidered the right pages at Apache and got a list of
3780 apache mailing lists to
send to LI. This is an interim measure until they implement an opt-out link
on their invitation emails.
But they kicked dev@hbase.a.o back at me saying that it was associated with
an account.
One nice feature is the ability to mark tests as skipped while still
reporting the skipped tests.
On Wed, Feb 23, 2011 at 10:45 AM, Ryan Rawson ryano...@gmail.com wrote:
I filed HBASE-3555, and I listed the following reasons;
- test groups allow us to separate slow/fast tests from each other
Also, ycsb at least doesn't check the data that is returned.
That means that their numbers for /dev/null would have been even better.
On Fri, Feb 11, 2011 at 7:36 AM, Jean-Daniel Cryans jdcry...@apache.orgwrote:
Finally, in research the most important step is validation of the
results. They
Online merge is a bit dangerous. Lots of applications require that the
table be set up pre-split. This is probably more common than the need for
merging.
Having such a pre-split table collapse before it is full would be a
disaster.
It should be pretty easy to script taking a few regions
See http://en.wikipedia.org/wiki/Double_hashing for information on double
hashing.
On Tue, Jan 25, 2011 at 8:11 AM, Nicolas Spiegelberg nspiegelb...@fb.comwrote:
A great article for Bloom Filter rules of thumb:
http://corte.si/posts/code/bloom-filter-rules-of-thumb/
Note that only rules #1
I was frustrated over the weekend using ycsb because it doesn't check the
data it gets and because of general code hygiene issues.
Rather than just kvetch, I have modified ycsb and pushed it back onto
github.
See https://github.com/tdunning/YCSB
My changes include:
a) switched to maven to
It is possible to do (but a pain) by setting up the smaller tasks as
sub-tasks on major bugs and then only setting the version for the major
bugs.
I don't like doing that so much since it is fairly intricate. If you have a
marker for majorness, you can probably build a custom report to do the
Nice work!
On Fri, Jan 21, 2011 at 12:40 PM, Mingjie Lai mingjie_...@trendmicro.comwrote:
Guys.
There is a discussion regarding testing HBASE with YCSB on Whirr or EC2.
Send to @dev so more people can be involved.
Lars.
I have an automatic YCSB test for HBase running on EC2. It was derived
Nice work Todd.
Were these numbers extracted using jconsole?
On Thu, Jan 20, 2011 at 9:33 AM, Todd Lipcon t...@cloudera.com wrote:
I did some experiments to understand our full GC issues better last night.
Here are the results:
http://people.apache.org/~todd/hbase-fragmentation/
Also, can you say what YCSB workload that used?
On Thu, Jan 20, 2011 at 9:33 AM, Todd Lipcon t...@cloudera.com wrote:
I did some experiments to understand our full GC issues better last night.
Here are the results:
http://people.apache.org/~todd/hbase-fragmentation/
/browse/HBASE-3455
https://issues.apache.org/jira/browse/HBASE-3455-Todd
On Thu, Jan 20, 2011 at 9:48 AM, Ted Dunning tdunn...@maprtech.com
wrote:
Also, can you say what YCSB workload that used?
On Thu, Jan 20, 2011 at 9:33 AM, Todd Lipcon t...@cloudera.com wrote:
I did some
This is on 0.90, right? Were you using HDFS to store your region tables?
I just ran into the same thing and looked into the
org.apache.hadoop.hbase.regionserver.wal.SequenceFileLogReader$WALReader$WALReaderFSDataInputStream.getPos
method.
That method does some truly hideous reflection things
On Fri, Jan 7, 2011 at 10:30 AM, Stack st...@duboce.net wrote:
As to your question Ted, it does seem like we could do the reflection
once-only in the constructor rather than every time we do a getPos.
Let me ask Nicolas. Maybe he had reason for having to do it each
time. As to its
I think a simple check for the presence of the method is better.
On Fri, Jan 7, 2011 at 11:32 AM, M. C. Srivas mcsri...@gmail.com wrote:
How about checking to see if in is instanceOf DFSInputStream before
doing
the rest of the stuff?
St.Ack
Great. I will file a patch to move the check to the constructor and fail
back to
old process if the method is missing.
For our case, I just implemented getFileLength and all is happy (on that
front)
On Fri, Jan 7, 2011 at 12:38 PM, Stack st...@duboce.net wrote:
Let me open an issue to add
If you have a PMC member who is willing to be release manager, what is the
beef?
On Thu, Dec 23, 2010 at 2:49 PM, Ryan Rawson ryano...@gmail.com wrote:
Looks like the fight does not go well. A lot of hdfs developers are
concerned that it would detract resources. I'm not sure who's
heaps would blunt the advantage of a C++ program using malloc.
-ryan
On Wed, Dec 15, 2010 at 11:15 AM, Ted Dunning
tdunn...@maprtech.com
wrote:
From the small comments I have heard, the RAM versus disk
difference is mostly what I have heard they were testing.
On Wed, Dec 15
45 matches
Mail list logo