if region A is doing compaction , load balancer ask region server to close
region A
if set hbase.hstore.close.check.interval = 0,
the region A can't be closed until the completion of compaction,
during this period, the region A can also accept mutation request?
Am I right?
Does this have bad
+1 on forwarding after 5 seconds.
On Thu, Jun 2, 2011 at 8:10 AM, Lars George lars.geo...@gmail.com wrote:
Hi,
When you click on split/compact in the table.jsp you get:
Split request accepted.
Reload.
Reloading is braindead as it triggers the action over and over again.
We should have a
Hi,
We have
pRegion names are made of the containing table's name, a comma,
the start key, a comma, and a randomly generated region id. To illustrate,
the region named
emdomains,apache.org,5464829424211263407/em is party to the table
emdomains/em, has an id of em5464829424211263407/em and the
Thanks Ted.
On Thu, Jun 2, 2011 at 4:19 PM, Anty anty@gmail.com wrote:
if region A is doing compaction , load balancer ask region server to close
region A
if set hbase.hstore.close.check.interval = 0,
the region A can't be closed until the completion of compaction,
during this period,
Lucene has a compact FST (Finite State Transducer) that's used for the
sorted terms index. I think this is the same type of functionality as
the HBase block index, eg, a sorted index of row ids? The FST is more
compact keeping every Nth row id in RAM. Does the HFile format allow
pluggable block
Currently BlockIndex is an inner class of HFile.
It would be nice to support pluggable block index implementations.
FYI
On Thu, Jun 2, 2011 at 9:09 AM, Jason Rutherglen jason.rutherg...@gmail.com
wrote:
Lucene has a compact FST (Finite State Transducer) that's used for the
sorted terms
It would be nice to support pluggable block index
implementations.
+1
Perhaps we do this in the scope of HFile v2?
https://issues.apache.org/jira/browse/HBASE-3857
- Andy
--- On Thu, 6/2/11, Ted Yu yuzhih...@gmail.com wrote:
From: Ted Yu yuzhih...@gmail.com
Subject: Re: Lucene's FST
The FST is more compact [than] keeping every Nth row id in RAM.
It would be nice to support pluggable block index implementations
Maybe we should try to support this prior to the HFile v2, which
instead uses a tree structure to layout the blocks? Eg, a pluggable
block index then becomes more
+1 on forwarding after 5 seconds. This part of the user interface is very
confusing.
-Rakesh
On Thu, Jun 2, 2011 at 5:30 AM, Joey Echeverria j...@cloudera.com wrote:
+1 on forwarding after 5 seconds.
On Thu, Jun 2, 2011 at 8:10 AM, Lars George lars.geo...@gmail.com wrote:
Hi,
When you
Make an issue Lars. Add to it the two +1s below!
St.Ack
On Thu, Jun 2, 2011 at 5:10 AM, Lars George lars.geo...@gmail.com wrote:
Hi,
When you click on split/compact in the table.jsp you get:
Split request accepted.
Reload.
Reloading is braindead as it triggers the action over and over
Hey Matt,
Interesting email, and also something I've been thinking about recently.
Unfortunately, I think one of the big prerequisites before we can start
thinking about actual compression algorithms is some refactoring around how
KeyValue is used. Currently, KeyValue exposes byte arrays in a
Memstore dynamic row key compaction
This is interesting though I think extremely difficult. We need this
for Lucene realtime search with the RAM terms dictionary, which is
slated to use a ConcurrentSkipListMap, eg it's lock free and it works.
As Todd mentions, the KeyValue's need for access to
I'm here!
Still parsing through the stuff in this e-mail but I agree that many approaches
will require rejiggering of KeyValue in a significant way. I have made some
attempts at this but nothing is very far. It's definitely something that we
are hoping to put some resources into over the
Hi,
I had developed a NOSQL Search HSearch
(http://bizosyshsearch.sourceforge.net/ ) where HBase is the data storage,
we indexed around 10 times of wikipedia information distributing in 10
Amazon EC2 machine and searching 100 concurrent users. It works wonderfully.
I presented my findings in
i agree about starting with the block index since it's a relatively
isolated piece of code. Could possibly be done without KeyValue
modifications
Right. I don't think any KeyValue changes will be required. I'm
trying to get more info about if we can MMap the FST, then store all
the keys
High-level this sounds like a great.
Inline below is some feedback and a bit of history on how we got here
in case it helps:
On Thu, Jun 2, 2011 at 3:28 PM, Matt Corgan mcor...@hotpads.com wrote:
* refer to prefix compression as compaction to avoid interfering with
traditional compression
On Thu, Jun 2, 2011 at 8:17 PM, Matt Corgan mcor...@hotpads.com wrote:
What about turning KeyValue into an interface with only the essential
getXX() methods? Whatever is using the KeyValue shouldn't *have* to know
what sort of structure it's backed by, but could figure out the
implementation
On Thu, Jun 2, 2011 at 9:51 PM, Stack st...@duboce.net wrote:
On Thu, Jun 2, 2011 at 8:17 PM, Matt Corgan mcor...@hotpads.com wrote:
What about turning KeyValue into an interface with only the essential
getXX() methods? Whatever is using the KeyValue shouldn't *have* to know
what sort of
18 matches
Mail list logo