But if we have two master in the master file we have master and secondary
node, *both *processes getting started on the two servers listed. Cant we
have master and secondary node started seperately on two machines??
On Fri, May 15, 2009 at 9:39 AM, jason hadoop wrote:
> I agree with billy. conf/m
Hi Andy,
Thanks for the tip.
I have a EC2 cluster with 6 nodes. Each a server grade large instance. I
have the mapred & regionservers running on all the nodes. Our deployment
will not go beyond 20 clusters in the near future. What would you suggest me
to have? Scenario 1 or 2 as u mentioned ?
On T
I like this a lot Guilherme. Perhaps we should open a JIRA with them so we
can track these great ideas.
On Thu, May 14, 2009 at 7:05 PM, Ryan Rawson wrote:
> Given the non core nature, I think the api should potentially facilitate
> this but the code should be contrib.
>
> On May 14, 2009 5:32 P
I agree with billy. conf/masters is misleading as the place for secondary
namenodes.
On Thu, May 14, 2009 at 8:38 PM, Billy Pearson
wrote:
> I thank the secondary namenode is set in the masters file in the conf
> folder
> misleading
>
> Billy
>
>
>
> "Rakhi Khatwani" wrote in message
> news:3848
I thank the secondary namenode is set in the masters file in the conf folder
misleading
Billy
"Rakhi Khatwani"
wrote in message
news:384813770905140603g4d552834gcef2db3028a00...@mail.gmail.com...
Hi,
I wanna set up a cluster of 5 nodes in such a way that
node1 - master
node2 - secondar
Given the non core nature, I think the api should potentially facilitate
this but the code should be contrib.
On May 14, 2009 5:32 PM, "Guilherme Germoglio" wrote:
On Thu, May 14, 2009 at 3:40 PM, stack wrote: > No
consideration has been made f...
I think so.
If nothing is to be changed on Row
On Thu, May 14, 2009 at 3:40 PM, stack wrote:
> No consideration has been made for changes in how locks are done in new
> 0.20.0 API. Want to propose something Guilherme? Could new zk-arbitrated
> locking be done inside the confines of the RowLock object?
I think so.
If nothing is to be chan
On Thu, May 14, 2009 at 1:07 PM, Joey Echeverria wrote:
> With a 5 server zk ensemble and an 80% write ratio, you should be able
> to support about 10,000 operations per second[1]. That sounds
> reasonable to me for most uses that require locks. If you require
> higher performance than that, then
I stand corrected, thanks for the info!
-ryan
On Thu, May 14, 2009 at 1:07 PM, Joey Echeverria wrote:
> With a 5 server zk ensemble and an 80% write ratio, you should be able
> to support about 10,000 operations per second[1]. That sounds
> reasonable to me for most uses that require locks. If
With a 5 server zk ensemble and an 80% write ratio, you should be able
to support about 10,000 operations per second[1]. That sounds
reasonable to me for most uses that require locks. If you require
higher performance than that, then locking probably isn't for you.
Taking advantage of versioning or
Ah I hate to be a cloud on a sunny day, but iirc, zk isn't designed for a
high write load. With thousands of requests a second one could overwhelm the
zk paxos consensus seeking protocol.
Another thing to remember is hbase doesn't "overwrite" values, it just
versions them. Perhaps this can be of h
No consideration has been made for changes in how locks are done in new
0.20.0 API. Want to propose something Guilherme? Could new zk-arbitrated
locking be done inside the confines of the RowLock object?
St.Ack
On Thu, May 14, 2009 at 9:44 AM, Guilherme Germoglio wrote:
> This way, HTable coul
No consideration has been made for changes in how locks are done in new
0.20.0 API. Want to propose something Guilherme? Could new zk-arbitrated
locking be done inside the confines of the RowLock object?
St.Ack
On Thu, May 14, 2009 at 9:44 AM, Guilherme Germoglio wrote:
> This way, HTable coul
Hi Alexandra,
Yes, a HBase release 0.19 is compatible with Hadoop 0.19 with no other
guarantees, and a HBase 0.20 release is compatible with Hadoop 0.20, and so on.
It is hard to say without actually trying a recompile and deployment if HBase
0.19.2 would have any trouble on Hadoop trunk. I wo
NEVER KILL -9 A REGION SERVER!!!
- Andy
From: Ninad Raut
To: hbase-user@hadoop.apache.org
Cc: Ranjit Nair
Sent: Thursday, May 14, 2009 3:03:06 AM
Subject: Re: Administration tool for HBase
Ryan,
Using bin/hbase-daemon.sh start regionserver and bin/hbase-d
Hi Ninad,
I think the answer depends on the anticipated scale of the deployment.
For small clusters (up to a few racks, ~40 servers per rack) I don't think
there is any significant performance hit to separate storage and computation.
Presumably all servers will share the same large GigE switch
This way, HTable could directly request for read or write row locks (
http://hadoop.apache.org/zookeeper/docs/current/recipes.html#Shared+Locks)
using zookeeper wrapper. The problem is that the client api would change a
little. Would these changes fit into the client api redesign for 0.20
(HBASE-12
On Wed, May 13, 2009 at 11:00 PM, Joey Echeverria wrote:
> Wouldn't it be better to implement the row locks using zookeeper?
>
THBase was done before ZK was in the mix. Now its here, we should look into
using it.
St.Ack
Hi,
I wanna set up a cluster of 5 nodes in such a way that
node1 - master
node2 - secondary namenode
node3 - slave
node4 - slave
node5 - slave
How do we go about that?
there is no property in hadoop-env where i can set the ip-address for
secondary name node.
if i set node-1 and node-2 in ma
i tried hbase 0.19.2 on hadoop 0.20.0 and had problems due to
versioning errors. 19 is v2 and 20 is v3 protocol or something along
those lines. so i went back down to hadoop 0.19.1 and hbase 0.19.2
and it all works fine
-sd
maybe i didn't do something right..
On Thu, May 14, 2009 at 1:16 PM, A
Hi,
Many thanks to all for your answers. Your help is much appreciated.
Andy, you were right, fs.trash.interval didn't help.However, on a second run
I was lucky enough that the overload got distributed over 2 of my 4
datanodes and it didn't result into an error during the import.
I was looking
Hi,
I am using Hadoop 0.19.1 and HBase 0.19.2.
I have ingested data in two tables. One of them 'observations' which is
expected to have a large number of records (for a medium size data set
10^9). Each record in table 'observations' refers to a certain star observed
which I store in a different
Dont kill database servers with kill -9 :-)
hbase-daemon.sh stop regionserver does a graceful shutdown, which can take a
while as data is flushed.
there was a bug, which i indicated, which can prevent root/meta from being
reassigned properly.
a few other notes:
- adding servers is easy, just sta
Ryan,
Using bin/hbase-daemon.sh start regionserver and bin/hbase-daemon.sh stop
regionserver can we add/remove slave nodes when the cluster is live?
How to handle a region not serving exception? Because in this scenario the
daemon.sh script seems to go on forever?
We usually kill the HRegion proces
Hi,
I want to get a design perspective here as to what will be the advantages of
seperating region servers and compute node(to run mapreduce tasks)
Will seperating datanodes from computes node reduce the load on the servers
and avoid swapping problems?
Will this seperation make map reduce tasks les
HBaseConfiguration config = new HBaseConfiguration();
config.set("hbase.master", "foo.bar.com:6");
On Thu, May 14, 2009 at 8:32 AM, monty123 wrote:
>
> Hi All,
>
> I am a newbie to hbase.
> I am able to setup hbase in pseudo-distributed mode and I have also done
> with its integration from
Hi All,
I am a newbie to hbase.
I am able to setup hbase in pseudo-distributed mode and I have also done
with its integration from Java. ( java client class and hbase were on same
system )
Now, I have no idea how to change configuration to access hbase from a
remote client ( like mysql jdbc conn
27 matches
Mail list logo