Question about developing mysql cluster

[nyaghma]

Hi,

I am Navid Yaghmazaeh, a Ph.D. student at the University of Texas at  
Austin. I'm member of a team working on a research project about  
adding some features to mysql cluster. A brief description of our  
project is that we are working on making mysql more durable by storing  
logs on disk at the end of each transaction. We want
to add this property to mysql in parallel to current property of  
making state of the system durable by epochs, so mysql cluster could  
support both level of durability at the same time.

In order to do so, we have to modify the Ndb code. I have some  
question about changing the mysql cluster code and if you could help  
me I would be greatly thankful.

Could you please link me to documents about protocols used in Ndb,  
specially in Ndb kernel, such as global checkpointing protocol or two  
phase commit protocol and protocols for committing transactions.? I  
have read the "Mysql Cluster API Developer Guide", but it is mostly  
about the Ndb API, and does not describe the Ndb kernel and the  
protocols used there.

Could please guide me where could be a good point to start modifying  
Ndb kernel code? As Ndb is a huge code and has many different modules  
which work together though they are separate in semantics, it would  
help a lot if you could help me about choosing a good starting point  
for applying changes.

Is there any available unit and integrity tests available in order to  
be able to test our changes and make sure everything is going well?  
How Could I test modules I modify and the functionality of the whole  
mysql cluster?

I would greatly appreciate your help in advance. Please do not  
hesitate to ask me to describe more if any thing about our work is not  
clear to you. I would be more than happy to discuss our work with you  
and hear your comments.

Thank you so much.
Regards,

Navid Yaghmazadeh
Ph.D. Student
Computer Science department
University of Texas at Austin