Is Gigaspace Cluster topology static ?

Hi All, As I know we specify the cluster topology & cluster members in the cluster-config.xml file. This file is read when the GSC is started. But if I want to change my cluster topology after I have created my Space, how do I do it ?

I am curious to understand if I can change the Gigaspace cluster topology dynamically ? Do I need to modify the cluster-config xml everytime the topology changes & Do I need to restart ?

Any light thrown on above questions would be much appreciated.

Thanks, Sudhindra

{quote}This thread was imported from the previous forum. For your reference, the original is [available here|http://forum.openspaces.org/thread.jspa?threadID=3678]{quote}

asked 2011-06-01 16:36:14 -0600

sudhindra_gk
1 ●1

updated 2013-08-08 09:52:00 -0600

jaissefsfex
1 ●1

edit retag flag offensive close merge delete

add a comment

Sudhindra,

In general , you should not use the cluster-config.xml file directly. It is an internal file used to generate the cluster information in runtime.
If you would like to deploy a clustered space and scale it to run across additional nodes you should use the elastic PU.
See:
http://www.gigaspaces.com/wiki/display/XAP8/ElasticProcessingUnit

You can run a simple example found below that is illustrating the dynamic scaling and the re-balancing of partitions across the GSCs without any system downtime:
http://www.gigaspaces.com/wiki/display/XAP8/ElasticProcessingUnit#ElasticProcessingUnit-EPUExample
Shay

answered 2011-06-01 16:43:10 -0600

shay hassidim

17818 ●8 ●12 ●16

edit flag offensive delete link

Comments

shay,

thanks a lot for the quick response. I did go over the link you mentioned. I had a few questions on that

1 In the section System BootStrapping it states that every system needs to start the Gigaspace Agent & the command line options instruct the agent to communicate with other. How does the agent communicate with other agents ? does it use the Locator or is based on the group name to find other agents ? if it is based on the group name then do we need to enable the multicast ?

2 The ESM does not support a redundant instance. If the system running ESM shuts down what happens to the cluster ?

3 Re-balancing Automatic Process Considerations it says - +During relocation of a specific instance, primary election takes place. For a few seconds, operations on that partition and operations on the whole cluster is denied. Internally, the client proxy retries the operation until the primary election takes place and masks the failure, but the delay exists.+ - Does this delay happen for every node/space relocation ?

4 The EPU works with Gigaspace version 8 but is there any solution available for gigaspace 7 ?

Any response will be really much appreciated.

Thanks, Sudhindra

sudhindra_gk ( 2011-06-02 14:53:03 -0600 )edit

Sudhindra, 1. agents communicate via the lookup service. They discover the lookup service using unicast (LOOKUPLOCATORS) or via multicast (LOOKUPGROUPS). 2. Starting with XAP 8.0.2 the ESM don't need a backup. It is totally stateless. If it fails it will be restarted automatically on one of the other machine running gs-agent. 3. During rebalancing , the backups instances are relocated first. In such a case this does not impact any clients. If a primary is relocated, it might impact existing clients accessing this primary, since these will be routed to use the relevant backup instance that turned to be a primary where the relocated primary turned into a backup in its new location. 4. Yes, there is a utility that performs similar activity as the ESM that works with XAP 7 (reblanacing). You will need to call it manually. You can find it here: http://www.gigaspaces.com/wiki/downlo...

Shay

shay hassidim ( 2011-06-02 16:16:49 -0600 )edit

add a comment

Is Gigaspace Cluster topology static ?

1 Answer

Comments

Your Answer

Question Tools

Stats

Related questions

Is Gigaspace Cluster topology static ? edit

1 Answer

Comments

Your Answer

Question Tools

Stats

Related questions

Is Gigaspace Cluster topology static ?