zookeeper-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jürgen Jakobitsch <jakobits...@punkt.at>
Subject Re: Zookeeper Directory Replication
Date Tue, 12 Apr 2011 16:02:47 GMT
hi ted,

thanks for answering..

i already found out that znodes are too small and actually not really suitable to store some
file data.

still i'm wondering what's being done with solr or neo4j or katta. 
for example what happens to a solr cluster when the master fails, if data is not replicated,
where does data come from?
if data is not replicated, what is then replicated?

also i was reading this [1] before, i might have understood you wrong, but i got it this way
: zookeeper coordinates, but storage 
is outside of zookeeper, nevertheless there are possibilities to get the whole thing up and
running.

wkr turnguard.com/turnguard


[1] http://zookeeper-user.578899.n2.nabble.com/is-zookeeper-suitable-for-my-application-td5614460.html

----- Original Message -----
From: "Ted Dunning" <ted.dunning@gmail.com>
To: user@zookeeper.apache.org
Cc: "Jürgen Jakobitsch" <jakobitschj@punkt.at>
Sent: Tuesday, April 12, 2011 5:22:49 PM
Subject: Re: Zookeeper Directory Replication

Zookeeper doesn't work the way that it seems that you think. 


Zookeeper stores data in znodes which have a directory-like naming scheme. Zookeeper does
not watch external files, nor does it provide you any way to access external files via the
Zookeeper API. 


In general, Zookeeper is used for synchronizing coordination information safely across a small
cluster. The information shared this way is safely replicated, but is intentionally limited
in size. The information is not even allowed to exceed available memory. 


On Tue, Apr 12, 2011 at 4:36 AM, Jürgen Jakobitsch < jakobitschj@punkt.at > wrote:



hi zookeepers, 

as a complete zk noob i'm really proud that i have managed to set up my first 
training cluster on two nodes that really works. 
unfortunately i'm not doing this for fun. i'm in need of synchronizing a whole directory 
between servers and know that it should be working with zk, since its working with solr 
and neo4j and they have a similar store to copy, it's only a little hard to grasp. 

could some be so kind as to point me into the right direction on how to use zookeeper to 
synchronize a whole directory with all its contents between nodes. i'm using openrdf sesame's

triple store, which stores it's datafiles in a single directory (most of these files 
get certainly bigger than 1MB) that i need to have available on different nodes. 


1. is this a suitable use case for zk? 
2. is there a standard way of doing this? 
3. how do i store a file (directory) reference in a znode and have that file (directory) watched
for changes? 

any hint 'n help really appreciated 
wkr turnguard.com/turnguard 

-- 
punkt. netServices 
______________________________ 
Jürgen Jakobitsch 
Codeography 

Lerchenfelder Gürtel 43 Top 5/2 
A - 1160 Wien 
Tel.: 01 / 897 41 22 - 29 
Fax: 01 / 897 41 22 - 22 

netServices http://www.punkt.at 

-- 
punkt. netServices
______________________________
Jürgen Jakobitsch
Codeography

Lerchenfelder Gürtel 43 Top 5/2
A - 1160 Wien
Tel.: 01 / 897 41 22 - 29
Fax: 01 / 897 41 22 - 22

netServices http://www.punkt.at


Mime
View raw message