Compared to pDPM-Central and Clover, pDPM-Direct does not need a centralized server for coordination. However, in order to keep a consistent running experience and to reduce development cost, we will run a server script similar to pDPM-Central and Clover. But this server will not run any pDPM-Direct component except memcached server instance.
Switch folder to pDPM/direct/herd/
.
Suppose we use three servers [S0, S1, and S2]
to run a [1 CN and 1 MN]
setting. We will run memcached on S0; a single MN on S1; and a single CN on S2. To start, run the following script at each server one by one:
- S0:
./run-servers.sh 0
to start memcached - S1:
./run-memory.sh 1
to start MN - S2:
./run-machine.sh 2
to start CN
Once experiment finishes, run pkill.sh
to terminate all relevant processes.
Most of the runtime configurations are controlled by bash scripts. We will use run-servers.sh
to demonstrate.
First, specify the memcached server IP. The script is able to start (and kill) memcached instance automatically. We should use the IP address of the coordinator.
export HRD_REGISTRY_IP="192.168.0.1"