Pacemaker clear failed actions automactically

I think the problem is in the init script, it does not respect the LSB spec.

If you look at the function haproxy_stop, in file /etc/init.d/haproxy:

haproxy_stop()
{
    if [ ! -f $PIDFILE ] ; then
        # This is a success according to LSB
        return 0
    fi
    for pid in $(cat $PIDFILE) ; do
        /bin/kill $pid || return 4
    done
    rm -f $PIDFILE
    return 0
}

In particularly, the line /bin/kill $pid || return 4. This makes the case that the process is killed the return value is 4, which according to the spec this is: user had insufficient privileges. Which is not correct.

In case of an error while processing any init-script action except for status, the init script shall print an error message and exit with a non-zero status code:

1 generic or unspecified error (current practice)
2 invalid or excess argument(s)
3 unimplemented feature (for example, "reload")
4 user had insufficient privilege
5 program is not installed
6 program is not configured
7 program is not running
8-99  reserved for future LSB use
100-149   reserved for distribution use
150-199   reserved for application use
200-254   reserved

You can try to change by:

/bin/kill $pid || return 7

the correct way is stop daemon with killproc(8) and if this fails killproc sets the return value according to LSB.

Eg.

/sbin/killproc -p $PIDFILE $HAPROXY

sends the signal SIGTERM to the pid found in $PIDFILE if and only if this pid belongs to $HAPROXY. If the named $PIDFILE does not exist, killproc assumes that the daemon of $HAPROXY is not running. The exit status is set to 0 for successfully delivering the default signals SIGTERM and SIGKILL otherwise to 7 if the program was not running. It is also successful if no signal was specified and no program was there for Termination because it is already terminated.

1- Is my assumption correct that on event 3, the returning node can be automatically joint to the cluster?

Yes, this can be done. DRBD should not go Primary on it's own unless told to in the resource configuration; check that the 'startup { become-primary-on }' definition is not set in the resource configs.

2- If it can be done, please tell how.

Check that the following are true:

a. 'drbd' is not set to start at boot ('chkconfig drbd off' in RHEL, 'update-rc.d drbd disable' in Debian).

b. DRBD should not be configured to become primary on it's own (as mentioned above).

The DRBD user's guide has a section on configuring DRBD for use with Pacemaker that might help if my answer above doesn't do the trick: https://drbd.linbit.com/users-guide/ch-pacemaker.html