• Project structure
• Compiling
• Running server processes
• Using client library
• Extended usage
• Running daemons manually

Project structure

We have one DB backend (named zoidberg) and one machine (bender), with server code running on it.

Compiling

At first, download latest version of cexpert to working directory. We need to extract it and prepare for building.

moriarty@bender:~/work/ce$ ls .
cexpert-0.8.tar.gz
moriarty@bender:~/work/ce$ tar -zxvf cexpert-0.8.tar.gz 
moriarty@bender:~/work/ce$ ls -al
total 328
drwxr-xr-x 3 moriarty users   4096 2009-04-21 15:44 .
drwxr-xr-x 4 moriarty users   4096 2009-04-21 15:43 ..
drwxr-xr-x 8 moriarty users   4096 2009-04-21 15:33 cexpert-0.8
-rw-r--r-- 1 moriarty users 315789 2009-04-21 15:42 cexpert-0.8.tar.gz
moriarty@bender:~/work/ce$ cd cexpert-0.8
moriarty@bender:~/work/ce/cexpert-0.8$ mkdir build
moriarty@bender:~/work/ce/cexpert-0.8$ cd build
moriarty@bender:~/work/ce/cexpert-0.8/build$ 

It's time to run standard building procedure:

moriarty@bender:~/work/ce/cexpert-0.8/build$ ../configure --enable-python=yes
moriarty@bender:~/work/ce/cexpert-0.8/build$ make && make check

If everything is ok, you should see something like that:

...........
make  check-TESTS
make[2]: Entering directory `/home/moriarty/work/ce/cexpert-0.8/build/tests'
shmem check: 0.1.2.3.4.5.6
shmem check: 0.1.2.3.4.5.6
DBChecker check 0
DBChecker check 1
DBChecker check 2
DBChecker check 3
DBChecker check 4
PASS: test_server
start client test
initiated, check now
PASS: test_cexpert_lib
==================
All 2 tests passed
==================
............

Now, run "sudo make install", which place libraries and headers to right places.

Running server processes

There are two ways of running daemons: using system-wide daemons set (via init.d script), or directly from client code (see Running daemons manually section).

Using init.d script:

moriarty@bender:~/work/ce/cexpert-0.8/build$/usr/local/bin/cexpert_s mysql://user:pasword@zoidberg:3306/test_test
moriarty@bender:~/work/ce/cexpert-0.8/build$ ps ux | grep cexpert_s
moriarty 26901  0.0  0.0  13892  1944 ?        Sl   15:58   0:00 /usr/local/bin/cexpert_s mysql.user.zoidberg.3306.test_test
moriarty 26914  0.0  0.0   2816   748 pts/9    S+   15:59   0:00 grep cexpert_s

You also can use init.d script:

moriarty@bender:~/work/ce/cexpert-0.8/build$ cat /etc/cexpert.d/test 
mysql://user:password@zoidberg:3306/test_test
moriarty@bender:~/work/ce/cexpert-0.8/build$ sudo /etc/init.d/cexpertd start
moriarty@bender:~/work/ce/cexpert-0.8/build$ ps aux | grep cexpert
root     26973  0.0  0.0  13756  1788 ?        Sl   16:01   0:00 /usr/local/bin/cexpert_s mysql.user.zoidberg.3306.test_test
moriarty 26993  0.0  0.0   2820   756 pts/9    S+   16:01   0:00 grep cexpert

Using client library

Python bindings is the easiest way to demonstrate usage (C and C++ libraries has similar interface).

moriarty@bender:~/work/ce/cexpert-0.8/build$ ipython
In [1]: import libpycexpert

In [2]: libpycexpert.get_is_available("mysql://user:password@zoidberg:3306/test_test")
Out[2]: 1

Now, let's broke server-DB connectivity:

moriarty@bender:~$ sudo iptables -A OUTPUT -p tcp --dport 3306 -j REJECT --reject-with tcp-reset

Check our python process:

In [3]: libpycexpert.get_is_available("mysql://user:password@zoidberg:3306/test_test")
Out[3]: 0

After restoring connectivity:

In [4]: libpycexpert.get_is_available("mysql://user:password@zoidberg:3306/test_test")
Out[4]: 1

Extended usage

It's possible to know when was the last time the daemon checks the status of the base:

In [17]: libpycexpert.get_is_available_ex("mysql://blogger:feedpass@zoidberg:3306/test3")
Out[17]: (1, 1241111982)
In [18]: libpycexpert.get_is_available_ex("mysql://blogger:feedpass@zoidberg:3306/test3")
Out[18]: (1, 1241112225)

Value, returned by get_is_available_ex is tuple (is_base_available, last_check_timestamp). Common way of using this values:

def get_db_status(connection_str):
   (is_ok, last_check_time) = libpycexpert.get_is_available_ex(connection_str)
   if time.time() - last_check_time > 5:
       #PANIC!!! Probably, the daemon is stopped. We could try to restart it, make a record in system log, or anything else. 
   return is_ok #last checked status

Running daemons manually

it's frequently use-case to connect by dynamically builded connection string. In such case, it can't be pre-written to /etc/cexpert.d/ files. So, we need capability to run daemons from our client code:

In [1]: import libpycexpert

In [2]: libpycexpert.is_registered("mysql://user:password@zoidberg:3306/test3")
Out[2]: 0

In [3]: libpycexpert.do_register("mysql://user:password@zoidberg:3306/test3")
Out[3]: 1

Now, the daemon starts:

moriarty@bender:~/src/cexpert$ ps ux | grep cexp
moriarty   622  0.0  0.0  13500  1676 ?        Sl   21:08   0:00 cexpert_s mysql.user.zoidberg.3306.test3

And we can check for DB's status or stop daemon:

In [7]: libpycexpert.get_is_available("mysql://user:password@zoidberg:3306/test3")
Out[7]: 1
In [8]: libpycexpert.do_unregister("mysql://user:password@zoidberg:3306/test3")
Out[8]: 1
In [9]: libpycexpert.is_registered("mysql://user:password@zoidberg:3306/test3")
Out[9]: 0