When building Data Store and Concurrent Data Store applications, the architecture decisions involve application startup (cleaning up any existing databases, the removal of any existing database environment and creation of a new environment), and handling system or application failure. "Cleaning up" databases involves removal and re-creation of the database, restoration from an archival copy and/or verification and optional salvage, as described in Handling failure in Data Store and Concurrent Data Store applications.
Data Store or Concurrent Data Store applications without database environments are single process, by definition. These applications should start up, re-create, restore, or verify and optionally salvage their databases and run until eventual exit or application or system failure. After system or application failure, that process can simply repeat this procedure. This document will not discuss the case of these applications further.
Otherwise, the first question of Data Store and Concurrent Data Store architecture is the cleaning up existing databases and the removal of existing database environments, and the subsequent creation of a new environment. For obvious reasons, the application must serialize the re-creation, restoration, or verification and optional salvage of its databases. Further, environment removal and creation must be single-threaded, that is, one thread of control (where a thread of control is either a true thread or a process) must remove and re-create the environment before any other thread of control can use the new environment. It may simplify matters that Berkeley DB serializes creation of the environment, so multiple threads of control attempting to create a environment will serialize behind a single creating thread.
Removing a database environment will first mark the environment as "failed", causing any threads of control still running in the environment to fail and return to the application. This feature allows applications to remove environments without concern for threads of control that might still be running in the removed environment.
One consideration in removing a database environment which may be in use by another thread, is the type of mutex being used by the Berkeley DB library. In the case of database environment failure when using test-and-set mutexes, threads of control waiting on a mutex when the environment is marked "failed" will quickly notice the failure and will return an error from the Berkeley DB API. In the case of environment failure when using blocking mutexes, where the underlying system mutex implementation does not unblock mutex waiters after the thread of control holding the mutex dies, threads waiting on a mutex when an environment is recovered might hang forever. Applications blocked on events (for example, an application blocked on a network socket or a GUI event) may also fail to notice environment recovery within a reasonable amount of time. Systems with such mutex implementations are rare, but do exist; applications on such systems should use an application architecture where the thread recovering the database environment can explicitly terminate any process using the failed environment, or configure Berkeley DB for test-and-set mutexes, or incorporate some form of long-running timer or watchdog process to wake or kill blocked processes should they block for too long.
Regardless, it makes little sense for multiple threads of control to simultaneously attempt to remove and re-create a environment, since the last one to run will remove all environments created by the threads of control that ran before it. However, for some few applications, it may make sense for applications to have a single thread of control that checks the existing databases and removes the environment, after which the application launches a number of processes, any of which are able to create the environment.
With respect to cleaning up existing databases, the database environment must be removed before the databases are cleaned up. Removing the environment causes any Berkeley DB library calls made by threads of control running in the failed environment to return failure to the application. Removing the database environment first ensures the threads of control in the old environment do not race with the threads of control cleaning up the databases, possibly overwriting them after the cleanup has finished. Where the application architecture and system permit, many applications kill all threads of control running in the failed database environment before removing the failed database environment, on general principles as well as to minimize overall system resource usage. It does not matter if the new environment is created before or after the databases are cleaned up.
After having dealt with database and database environment recovery after failure, the next issue to manage is application failure. As described in Handling failure in Data Store and Concurrent Data Store applications, when a thread of control in a Data Store or Concurrent Data Store application fails, it may exit holding data structure mutexes or logical database locks. These mutexes and locks must be released to avoid the remaining threads of control hanging behind the failed thread of control's mutexes or locks.
There are three common ways to architect Berkeley DB Data Store and Concurrent Data Store applications. The one chosen is usually based on whether or not the application is comprised of a single process or group of processes descended from a single process (for example, a server started when the system first boots), or if the application is comprised of unrelated processes (for example, processes started by web connections or users logging into the system).
When this process starts, it removes any existing database environment and creates a new environment. It then cleans up the databases and opens those databases in the environment. The application can subsequently create new threads of control as it chooses. Those threads of control can either share already open Berkeley DB DB_ENV and DB handles, or create their own. In this architecture, databases are rarely opened or closed when more than a single thread of control is running; that is, they are opened when only a single thread is running, and closed after all threads but one have exited. The last thread of control to exit closes the databases and the database environment.
This architecture is simplest to implement because thread serialization is easy and failure detection does not require monitoring multiple processes.
If the application's thread model allows the process to continue after thread failure, the DB_ENV->failchk method can be used to determine if the database environment is usable after the failure. If the application does not call DB_ENV->failchk, or DB_ENV->failchk returns DB_RUNRECOVERY, the application must behave as if there has been a system failure, removing the environment and creating a new environment, and cleaning up any databases it wants to continue to use. Once these actions have been taken, other threads of control can continue (as long as all existing Berkeley DB handles are first discarded), or restarted.
This architecture requires the order in which threads of control are created be controlled to serialize database environment removal and creation, and database cleanup.
In addition, this architecture requires that threads of control be monitored. If any thread of control exits with open Berkeley DB handles, the application may call the DB_ENV->failchk method to determine if the database environment is usable after the exit. If the application does not call DB_ENV->failchk, or DB_ENV->failchk returns DB_RUNRECOVERY, the application must behave as if there has been a system failure, removing the environment and creating a new environment, and cleaning up any databases it wants to continue to use. Once these actions have been taken, other threads of control can continue (as long as all existing Berkeley DB handles are first discarded), or restarted.
The easiest way to structure groups of related processes is to first create a single "watcher" process (often a script) that starts when the system first boots, removes and creates the database environment, cleans up the databases and then creates the processes or threads that will actually perform work. The initial thread has no further responsibilities other than to wait on the threads of control it has started, to ensure none of them unexpectedly exit. If a thread of control exits, the watcher process optionally calls the DB_ENV->failchk method. If the application does not call DB_ENV->failchk or if DB_ENV->failchk returns DB_RUNRECOVERY, the environment can no longer be used, the watcher kills all of the threads of control using the failed environment, cleans up, and starts new threads of control to perform work.
One solution is to log a thread of control ID when a new Berkeley DB handle is opened. For example, an initial "watcher" process could open/create the database environment, clean up the databases and then create a sentinel file. Any "worker" process wanting to use the environment would check for the sentinel file. If the sentinel file does not exist, the worker would fail or wait for the sentinel file to be created. Once the sentinel file exists, the worker would register its process ID with the watcher (via shared memory, IPC or some other registry mechanism), and then the worker would open its DB_ENV handles and proceed. When the worker finishes using the environment, it would unregister its process ID with the watcher. The watcher periodically checks to ensure that no worker has failed while using the environment. If a worker fails while using the environment, the watcher removes the sentinel file, kills all of the workers currently using the environment, cleans up the environment and databases, and finally creates a new sentinel file.
The weakness of this approach is that, on some systems, it is difficult to determine if an unrelated process is still running. For example, POSIX systems generally disallow sending signals to unrelated processes. The trick to monitoring unrelated processes is to find a system resource held by the process that will be modified if the process dies. On POSIX systems, flock- or fcntl-style locking will work, as will LockFile on Windows systems. Other systems may have to use other process-related information such as file reference counts or modification times. In the worst case, threads of control can be required to periodically re-register with the watcher process: if the watcher has not heard from a thread of control in a specified period of time, the watcher will take action, cleaning up the environment.
If it is not practical to monitor the processes sharing a database environment, it may be possible to monitor the environment to detect if a thread of control has failed holding open Berkeley DB handles. This would be done by having a "watcher" process periodically call the DB_ENV->failchk method. If DB_ENV->failchk returns DB_RUNRECOVERY, the watcher would then take action, cleaning up the environment.
The weakness of this approach is that all threads of control using the environment must specify an "ID" function and an "is-alive" function using the DB_ENV->set_thread_id method. (In other words, the Berkeley DB library must be able to assign a unique ID to each thread of control, and additionally determine if the thread of control is still running. It can be difficult to portably provide that information in applications using a variety of different programming languages and running on a variety of different platforms.)
Obviously, when implementing a process to monitor other threads of control, it is important the watcher process' code be as simple and well-tested as possible, because the application may hang if it fails.
Copyright (c) 1996-2005 Sleepycat Software, Inc. - All rights reserved.