A Protocol to Detect and Kill Orphan Processes in Distributed Computer Systems

The protocol presented here is to detect and kill orphan processes [1] in distributed computer systems effectively. That is it detects and kills the orphans when they are born. It does not waste any time to kill the orphans as the previous approaches do. All Remote procedure calls (RPC) [21][3][4] are recorded in the global log[5][6] which is associated with the monitor process[5][6] which is kept in the system to watch constantly the RPCs to find out the orphan computations. It is done by the monitor process which sends a token to those clients who participates in the RPCs and find out the active clients and abort [7] or crash processes [7]. The protocol is very much effective in the sense that the orphans are killed immediately after their birth. This protocol also handles the nested invocation [8] of RPCs. So, this protocol maintains data consistency [9] in the system and it avoids the wastage of valuable computer resources. All other approaches wait until the abort or crash process gets rebooted. Till the crash processes get rebooted, the orphans are active and make unwanted result which may lead to inconsistency [10] of data and moreover the wastages of valuable resources like CPU, memory and database etc.