open-chord-users

[Open-chord-users] problems running open chord on planetlab

[Arne R. <arn...@rw...>]

Hello!

first of all, thanks to the creators of this fine piece of software!

I'm trying to deploy open chord on planetlab (basically just distributed 
computer systems. url: planet-lab.org).
I'm using the socket-protocol and only managed to achieve a chord-ring 
size  of about 100 peers. Already before this size, strange things seem 
to occur. Most of the times, a new peer is unable to bootstrap, because 
he can't seem to create a socket to the bootstrapping-peer or some other 
peer. Other times, the joining peer waits for a reply until the end of 
time (poetically speaking). Right now I'm trying to use the rmi-protocol 
  without success as well. One of the things I noticed there, were peers 
kicking (still active) successors out of there references.

I'm using "Open Chord 1.0.5" and had no problems running it using the 
local-protocol or the socket-protocol on some virtual machines on one 
computer.

Has anyone a clue or experience in deploying OC in a realistic scenario 
with a size above 100 peers?

Thanks in advance and best regards,
     Arne

Re: [Open-chord-users] problems running open chord on planetlab

[Cesar A. <ces...@gm...>]

Hi,

I tried with 10 machines using RMI protocol, and works fine.
I can't see a plausible reason to this happen if your configuration file
admit
this number of nodes.
Seems you are getting problems with different machines (remote nodes), check
your
security settings:

java -Djava.security.policy=policyfilename
policyfilename
<<
grant {
  permission java.security.AllPermission;
};
>>

You can try, increase the value of timeout, just for foil any load problem.

On Tue, May 19, 2009 at 3:42 PM, Arne Rache <arn...@rw...>wrote:

> Hello!
>
> first of all, thanks to the creators of this fine piece of software!
>
> I'm trying to deploy open chord on planetlab (basically just distributed
> computer systems. url: planet-lab.org).
> I'm using the socket-protocol and only managed to achieve a chord-ring
> size  of about 100 peers. Already before this size, strange things seem
> to occur. Most of the times, a new peer is unable to bootstrap, because
> he can't seem to create a socket to the bootstrapping-peer or some other
> peer. Other times, the joining peer waits for a reply until the end of
> time (poetically speaking). Right now I'm trying to use the rmi-protocol
>  without success as well. One of the things I noticed there, were peers
> kicking (still active) successors out of there references.
>
> I'm using "Open Chord 1.0.5" and had no problems running it using the
> local-protocol or the socket-protocol on some virtual machines on one
> computer.
>
> Has anyone a clue or experience in deploying OC in a realistic scenario
> with a size above 100 peers?
>
> Thanks in advance and best regards,
>     Arne
>
>
>
> ------------------------------------------------------------------------------
> Crystal Reports - New Free Runtime and 30 Day Trial
> Check out the new simplified licensing option that enables
> unlimited royalty-free distribution of the report engine
> for externally facing server and web deployment.
> http://p.sf.net/sfu/businessobjects
> _______________________________________________
> Open-chord-users mailing list
> Ope...@li...
> https://lists.sourceforge.net/lists/listinfo/open-chord-users
>