Even though the explanation on the website is already good (client-replica interaction), I felt that, like in the course book, the full algorithm written out in text form can still be a bit hard to follow. When I was studying for the exam, I wrote it out as pseudocode, and that helped me a lot.

The pseudocode is entirely based on the course book (pages 785-791), and I tried to follow the same style used for other algorithms in the book.

I split it into four parts to make it easier to read and understand:

• Envio de pedido de leitura (query) do cliente para a réplica
• Envio de pedido de modificação (update) do cliente para a réplica
• Troca de mensagens de gossip entre réplicas
• Operação de merge entre vector timestamps

Hope this is helpful.