We would be moving to a clustered environment from a standalone mode. I would greatly appreciate if some pointers/guidelines are available as to what coding practices to be followed. Or rather what pitfalls need to be avoided so that when we move to the clustered environment, the existing code would not break.
One example, is if the flow services are reading/writing to local directories, it will break when we move to a clustered env, if those directories are not present on say the other IS node.
I am looking for a comprehensive list or a checklist of such things to be taken care/reviewed in the exisitng code to ensure that it does not break when moved to clustered environment.
Thanks in advance,
I am not sure about you go the reply for the same but here are some points which can be helpful:
Take care of your scehdulers, it might not be opted for scheduled in cluster option at the time of stand alone mode
Take care of publishable document types. Make sure not more than one broker document tyoe is created and all the nodes int he cluster have the same broker document type
Take care of triggers…make sure they are not modified in cluster mode. All the modifications to the trigger usually shoudl be done in stand alone server and deployed to cluster instead of modifying them on cluster itself
Take care of file writing operations which may write them to separate locations
Do not use save/restore pipelines in cluster mode…basically in a pub-sub arch u dont know which server willl subscribe to the document
Make sure you use file.sparators while coding services which frame your file names and all those stuff…so that change of underlying OS does not affect your code
Make sure when u use services like getServerName, you include port too as different servers in cluster can have the same ip but different ports