OK, I found the reason.
The task queue for ROW-READ-STAGE ThreadPoolExecutor is a Bounded queue, and
its size is 4096 (defined by DatabaseDescriptor.getStageQueueSize())
When the pending tasks of ROW-READ-STAGE ThreadPoolExecutor is more than
4096, the follow-up submitted task will be
Hi:
These days I found my Cassandra is strange, much slower than before. And
I Spent much time to figure it out and today I got the answer.
Some bad buy keeps on writing many data day and night, then made a very big
row mutation which size is about 140M.
In this period I restarted
Drew,
I was wondering if you care to share your map-reduce code
Thanks
Carlos
From: Drew Dahlke [drew.dah...@bronto.com]
Sent: Monday, June 28, 2010 7:17 AM
To: user@cassandra.apache.org
Subject: Re: Map Reduce support
The difference is noticeable but
I'm afraid I didn't hold on to it, sorry folks
On Mon, Jun 28, 2010 at 8:58 AM, Carlos Sanchez
carlos.sanc...@riskmetrics.com wrote:
Drew,
I was wondering if you care to share your map-reduce code
Thanks
Carlos
From: Drew Dahlke
On yesterday's This Week in Techhttp://www.twit.tv/254 (TWiT) podcast with
Leo Laporte (Wiki: http://wiki.twit.tv/wiki/TWiT_254), Kevin Rose of
Digghttp://digg.com/ fame was a guest. He gave a public preview of the new
Digg 4; it looks very nice and should be released in the next month or two.
Yes, you should increase your timeout if you are hinting big mutations
(or big rows that were built from smaller mutations).
2010/6/28 Lu Ming xl...@live.com:
Hi:
These days I found my Cassandra is strange, much slower than before.
And I Spent much time to figure it out and today I got
The principle in Cassandra modeling is that for each query, you should
denormalize your data at write time such that you can get the data for
that query from a single row.
The best explanation so far is at
http://www.rackspacecloud.com/blog/2010/05/12/cassandra-by-example/
On Tue, Jun 22, 2010
If you're getting to where you have thousands of backed-up commands on
RRS then it doesn't really matter what executor behavior is, because
you're severely screwed. :)
Often this means you are i/o bound and you should look into row or key
caching. Possibly you may simply need to add capacity to
Yes.
On Sat, Jun 26, 2010 at 12:53 AM, Todd Nine t...@spidertracks.co.nz wrote:
Hi all,
I'm having a lot of problems getting Lucandra to correctly handle numeric
document fields. After examining the keys it has written to the CF, I
believe it may be an issue of column ordering by bytes.
On 6/28/10 3:29 AM, Lu Ming wrote:
Every one hour HintedHandOffManager will check hintedhandoff
ColumnFamily then send out the big rowmutations to alive nodes,
It fails again because of the TimeoutException, so the task will never
finish and the big rowmutation is sending again and again.
In
Digg is not forking Cassandra. We use 0.6 for production, with a few in-house
patches (related to our infrastructure). The biggest difference with our branch
and apache 0.6 branch is we have the work Kelvin and Twitter has done in
regards to Vector Clocks + Distributed Counters. This will never
If you're interested:
https://issues.apache.org/jira/browse/CASSANDRA-1072
https://issues.apache.org/jira/browse/CASSANDRA-580
-Kelvin
On Mon, Jun 28, 2010 at 9:35 AM, Chris Goffinet c...@chrisgoffinet.com wrote:
Digg is not forking Cassandra. We use 0.6 for production, with a few
in-house
The approach that I took incorporates some of the ideas listed here...
Basically each message in the queue was assigned a sequence number (needs to
be unique and increasing per queue), and then read out in sequence number
order.
The Message CF is logically one row per queue, with each column being
Carlos Alvarez cbalvarez at gmail.com writes:
On Sat, Apr 24, 2010 at 10:59 PM, Jonathan Ellis jbellis at gmail.com
wrote:
No, Framed is totally different.
You are right. Seeing both, the java and c# thrift code, I think that
there is no need to use other transport than TSocket in Java
Hi everyone,
Can anyone point me to an in-depth description of clustertool
functionality?
Specifically interested in the global_snaphot/clear_global_snapshot options.
The terse --help doesn't tell much.
Thanks in advance!
Carlos
First let me premise, I am new to Cassandra. I just got it installed
and was able to add data, connect to Cassandra via Thrift and retrieve
the data. Since, Thrift uses RPC, I was wondering if Cassandra uses
Thrift on the server side to handle the requests from the clients? I
know Thrift is
On Mon, Jun 28, 2010 at 1:30 PM, Peter Minearo
peter.mine...@reardencommerce.com wrote:
First let me premise, I am new to Cassandra. I just got it installed and
was able to add data, connect to Cassandra via Thrift and retrieve the
data. Since, Thrift uses RPC, I was wondering if Cassandra
On Mon, Jun 28, 2010 at 9:35 AM, Chris Goffinet c...@chrisgoffinet.com wrote:
Digg is not forking Cassandra. We use 0.6 for production, with a few
in-house patches (related to our infrastructure). The biggest difference
with our branch and apache 0.6 branch is we have the work Kelvin and
Would it ever be useful to someday have browser clients access cassandra
servers directly? I imagine that would be the most compelling scenario to
have REST API for.
On Mon, Jun 28, 2010 at 1:36 PM, Paul Prescod p...@prescod.net wrote:
On Mon, Jun 28, 2010 at 1:30 PM, Peter Minearo
You answered my questions on that. Thanks!!
-Original Message-
From: pres...@gmail.com [mailto:pres...@gmail.com] On Behalf Of Paul Prescod
Sent: Monday, June 28, 2010 1:37 PM
To: user@cassandra.apache.org
Subject: Re: Cassandra and Thrift on the Server Side
On Mon, Jun 28, 2010 at
On Mon, Jun 28, 2010 at 1:45 PM, Marty Greenia martygree...@gmail.com wrote:
Would it ever be useful to someday have browser clients access cassandra
servers directly? I imagine that would be the most compelling scenario to
have REST API for.
This is an interesting idea, but introduces quite a
On Mon, 2010-06-28 at 13:45 -0700, Marty Greenia wrote:
Would it ever be useful to someday have browser clients access
cassandra servers directly? I imagine that would be the most
compelling scenario to have REST API for.
I think thats an interesting use case for REST.
From what I've seen,
I agree, it would probably make more sense to just use a conventional http
server to interface with the browser clients on the front-end to act as a
pass-through to cassandra on the back-end. No sense re-implementing all that
functionality.
Still, to Clint's point, everyone knows how to make an
Joe,
(Disclaimer: I work here) Check out http://www.simplegeo.com, we're
building a horizontally scalable spatial database on top of Cassandra
along with geographic analytics, data provisioning, and hosting
services. Using us might be easier than hacking around in the weeds
with MongoDB.
Sorry
So no one is able to help?
From: Anthony Ikeda [mailto:anthony.ik...@cardlink.com.au]
Sent: Monday, 28 June 2010 12:36 PM
To: user@cassandra.apache.org
Subject: Understanding SuperColumns
I seem to be scratching my head about how to model Super Columns
properly and how they relate to a 1
Clustertool is nodetool for cluster-wide operations.
On Mon, Jun 28, 2010 at 2:27 PM, Carlos Armas carl...@gmail.com wrote:
Hi everyone,
Can anyone point me to an in-depth description of clustertool
functionality?
Specifically interested in the global_snaphot/clear_global_snapshot options.
Hi All,
We have a Cassandra cluster setup with 2 EC2 m1.large machines, one is
running in us-east zone whereas 2nd one is running in us-west zone. As part
of Seed section, I provided each other's node's EC2 public DNS entry in each
of storage-conf.xml. but when I ran nodetool command, I am not
What are you using for ListenAddress and ThriftAddress in the
storage-conf.xml file? By default, they are set to 'localhost', which means
Node A is probably telling Node B to contact itself on the private IP
address.
Marty
On Mon, Jun 28, 2010 at 7:18 PM, Lenin Gali galile...@gmail.com wrote:
On Sun, Jun 27, 2010 at 7:36 PM, Anthony Ikeda
anthony.ik...@cardlink.com.au wrote:
Say my query is: Get all Work addresses in New York and the address
owner. Steps to get the data would be:
If this is the query you want to run, then you probably just want to put the
owner in the index
Hi, all
We have several nodes in DC1 and DC2 and we want to move all nodes in DC2 to
a new DC3, also IPs will be changed. The whole progress will last about 1-2
days.
Can we simply stop all nodes in DC2, move to DC3, chage IP config in
storage-conf.xml then start all nodes in DC3?
Or shall we do
I have provided East node's public DNS/ip address at that both end points
( ListenAddress and ThriftAddress) and West instance's public IP address added
as part of Seed for east.
Similarly I used West node's public DNS/IP for ListenAddress and ThriftAddress
and included East node's public IP
Thanks Ben. I've just been trying to nail this down - it's difficult
knowing if you are getting it right when it feels like you're going
backwards.
Anthony
From: Benjamin Black [mailto:b...@b3k.us]
Sent: Tuesday, 29 June 2010 1:00 PM
To: user@cassandra.apache.org
Subject: Re:
now i am confused about using the operation batch_mutate,when i follow the code
from
here:http://stackoverflow.com/questions/2783323/cassandra-batch-mutate/3137753#3137753
and it seems it has success inserted something,but when i use cassandra-cli to
check it,it just return nothing,
could
33 matches
Mail list logo