Re: mysql 8 NDB php connections drop, cant connect
To be a lot clearer, its not a problem with configuration from php. I'm using a common platform called moodle and the end goal for deployment is that the final goal for deployment must support 2500 user actions per second (setup i currently uses need to at least be able to support 500 per second).
At the moment the issue i have is with the cluster as i'm trying to wrap my head around it. nginx or php failure does not impact usage, but at the moment i'm still figuring out mysql.
i am using the latest stable mysql 8 cluster. 1 mysql control node, 2 nodes running both data and mysql (API). Whenever i hit any mysql node, it rejects anymore connections when the NDBD process is using a full CPU core despite many more cores available.
The 2nd data/mysql node doesn't accept any connections, so right now i have to set as static for php to use node1, but i want to load balance between the 2. When i try to do so, all the connections to node 2 fails, even if i set it staticly on the same network.
I'm trying to find if there is anything i missed or a setting i set differently. I've not found any dedicated configuration files for ndb data nodes, only mysql (stock) and mysql control.
I have many ways to load balance but first i have to make sure the nodes work and that they do not reject more connections when ndbd is fully using 1 CPU core and not using anymore cores.
The only bottleneck i see is ndbd process using 1 core only. mysql, php, nginx, redis don't even use the full CPU, not even a full core.