Configuring parameters for server clustering

Previous Member

Hi All,

 

I am trying to configure server clustering using the instructions here (Cluster ) I am having some issues with the configuration. For anyone that has been successful, could you share your configuration? My goal is to set up the clustering for data loading (so that the slave loads data while the master is accessed by users.) I have configured the AdulaParamsMaster and AdulaParamsSlave as specified in the doc, but it does not appear that they are interacting,and i am not sure how to confirm. When i start the master server, should the slave automatically start as well (or vice versa?)

 

Any assistance would be appreciated.

Björn Reuber

Hi,

 

for a complex setup like a Cluster I would allways recommend to have your local BOARD Subsidary included in the Project to help you with configuration.

 

So you have 2 different machine, running board, one for Master (and Slave 1) and one for Slave 2, correct?

 

Can you please post your adula configs, cause the master is quite easy, but for the slave's it does need more configuration.

 

Furthermore, should they have a "shared" BOARD folder, or has every slave instance a own BOARD folder? Do you want to use it for one DB or more? How many dbs are hosted on the server,.... without knowing these key facts it's very hard to help you

 

regards

Björn

Björn Reuber

Hi,

 

as a further hint, in the past we had some projects, were we used a Cluster for DataLoading. Mostly this was done, cause the DR were blocking events in older BOARD Version. On new Projects I would suggest to use 10.3.1 Version with HC Engine (DR on a single DB can run in parallel). So I would suggest to get in touch with your local BOARD subsidary to discuss, if Cluster or new BOARD versions would be the best approach for you

 

regards

Björn

Previous Member

Hi Bjorn,

 

Thanks for your reply. I understand your point about 10.3.1, but not quite ready to adopt it yet, so would like to try clustering as it also alleviates issues with post-load processing, etc. To your points above, yes, one master and one slave, using a shared board folder, and would like to use for 1 db for now (although there are multiple dbs.) My question is around the adulaparamsslave file, as what is unclear on the help site is if the information for all the slaves goes in the same file that is then put on all the slave machines (using multiple <SlaveDataType> tags) or if each AdulaParamsSlave file is distinct (and only has 1 <SlaveDataType> tag.) If you can help me with this i think i will be going in the right direction; if I still cannot get it I will log a ticket

Björn Reuber

Hi,

 

every slave instance will have a own configuration file, cause of SlaveName (can be choosen), Slave Adress.

I would also suggest to have the different slaves running on different Machines (so not having different slaves running on the same machine), cause in first case its easier to understand how it is working

 

regards

Björn