Client server application handling hospital patient data. Part of an undergrad course homework.
Consists of the following processes:
- whoServer
- whoClient
- master
- worker
The master process first forks a number worker processes and distributes to them the task of reading all the input files. The patient data are then sent to and stored by the whoServer process in appopriate data structures.
whoClient sends queries to whoServer, then whoServer forwards these queries to the worker processes if needed, waits for their response, and finally sends the response back to whoClient.
whoServer is a multi threaded process which assigns each new incoming connection to a thread. whoClient is also a multi threaded process which assigns each query to a thread.
The graph below depicts the interactions between the processes.
whoServer runs first. Sets up sockets and listens on both ports assigned. Creates threads, and enters "server mode", meaning waiting for a new connection and when arrived, places fd in pool. Threads wait to get a file descriptor.
master runs second. Reads input directory, forks workers and assigns them countries. Then pauses, waiting for a signal. On SIGINT or SIGTSTP, he sends SIGINT to workers, prompting them to exit, waits for them and then exits himself.
worker connects to whoServer, reads countries assigned, and sends summary statistics to whoServer. Then waits on receive_id(), meaning waiting in case whoServer needs to answer to a /searchPatientRecord command. That is the only command in which whoServer communicates with workers. All others are answered using the summary statistics. This was my design in hw2 and I kept it the same way.
When whoServer receives a new connection from statisticsPortNum, a thread picks it up, updates global data structures (same as hw2), and then stores file descriptor corresponding to the worker connection. That way the connection is not closed and when whoServer needs to communicate with workers, he iterates over the set of worker file descriptors.
whoClient runs next assigning each thread a line from queries.txt. Note that the parameter numThreads must be less or equal to the number of lines in queries.txt. A thread will connect to whoServer, send the query assigned to whoServer, wait for the answer and then print them both.
whoServer exits normally on SIGINT or SIGTSTP, meaning freeing apporpiately all resources.
The graph below illustrates the basic design idea for storing the patient data, which was chosen based on the nature of the queries we had to process. Note that the use of the C++ STL was not allowed and we had to create all structs we had to use on our own (see /include/mySTL/).
From the root project directory, type make.
Run the following commands from the root project directory. For additional info on the parameters of each process, execute it without any arguements.
- First create the input folder.
./scripts/create_infiles.sh ./scripts_input/diseases.txt ./scripts_input/countries.txt input_dir 5 8
- Then get whoServer up and running with
./bin/whoServer -q 8000 -s 9000 -w 5 -b 20
- Then in a new terminal
./bin/master -w 4 -b 64 -s 127.0.0.1 -p 9000 -i input_dir/
- And finally in another terminal
./bin/whoClient -q queries.txt -w 8 -sp 8000 -sip 127.0.0.1
- For date files in the
dd-mm-yyyyor the(d)d-(m)m-yyyyformat, set the appropriate macro at the begining ofinclude/myDate.hfile. By default, the(d)d-(m)m-yyyyis used, because the scriptcreate_infiles.shcreates date files in that format. - The serverIP parameter is in dotted decimal notation.