I have a server that holds all the data, and two nodes. The nodes have a similar table to the server that grab data from it and mark that it grabbed the data, so you do a simple
"select * from table where status='new'". from the node to the servers MYSQL. So all the nodes call this and grab new data. This is done using PHP.
from the node to the server (So we are doing remote MYSQL grabbing here of records.)
Then once it grabs it, i t changes the status to 'complete' on the main server. This was meant to prevent another node from grabbing the same data however because of latency sometimes the nodes grab the same record and process it. Both marking it as complete. It is not common but it does happen. So it is checking at about the same time, processing it because the record hasn't been marked 'complete' by the node until it's actually grabbed it.
So picture two node servers grabbing from the main server, at the same time, running the select * where status=new query, and processing/marking the record as status=complete, at similar times, enough that the latency is doing that.
How can I ensure that the nodes aren't grabbing the same records? Does anyone have any idea? I want to be able to use for example 10 node servers, not just 2 if needed without issues. At this time I think it could cause alot of duplicate grabbing and issues through MYSQL.
I understand that the nodes grabbing the data could be a problem. it would prevent it if the main server had a script and handled it, but the nodes are meant to be pluggable without any interaction or changes on the main server side (telling it that theres a new node etc isn't needed.) Few other things, but yeah, script is on the node sides ;)
Thanks guys, any advice is helpful. Even if it reduces it a large amount.