Arbortext PE 子进程 Allocation
There are several ways to allocate an Arbortext PE 子进程. The most common way to allocate an Arbortext PE 子进程 follows:
1. A dynamic component makes a call to the com.arbortext.e3.E3RequestContext.allocateE3Subprocess method; the call takes the HTTP request as a parameter. The allocateE3Subprocess method queries every configured Arbortext PE 子进程 pool, calling each pool's com.arbortext.e3.E3SubprocessPool.testRequest method. The testRequest determines whether the Arbortext PE 子进程 in that pool are supposed to handle requests like the one specified in the HTTP request.
If the response is true, the subprocess pool attempts to allocate an Arbortext PE 子进程 to service the request according to the next steps.
If the answer is false, allocateE3Subprocess continues to query the next Arbortext PE 子进程 pool in the list.
2. If a true response is returned for testRequest, then allocateE3Subprocess calls the Arbortext PE 子进程 pool's allocate method. The allocate method eventually either returns an Arbortext PE 子进程 object or throws an exception after processing the following:
a. The Arbortext PE 子进程 pool begins the allocation process by checking for any idle Arbortext PE 子进程. If so, the one idle for the shortest time is allocated. If not, the Arbortext PE 子进程 pool checks whether the configured maximum number of Arbortext PE 子进程 are running. If not, it starts a new one and, when it is ready to accept requests, returns the Arbortext PE 子进程 object.
b. If all Arbortext PE 子进程 in the pool are busy and the pool is configured to cascade to another pool, allocate returns the result of the cascaded pool's allocate method.
c. If all
Arbortext PE 子进程 in a pool (and its cascaded pools) are busy and no additional ones can be started, the
allocate method waits until either an
Arbortext PE 子进程 is no longer busy or the waiting period expires (explained in
The maxSubprocessWait Attribute). If an
Arbortext PE 子进程 is freed during the waiting period,
allocate returns its object.
3. If, after this processing, no Arbortext PE 子进程 are found, allocate throws an exception and transmits an All Arbortext PE 子进程 are currently busy error.
Arbortext PE 子进程 are allocated to process queued transactions by the Queued Transaction Scheduler.
1. The Queued Transaction Scheduler begins by asking an
Arbortext PE 子进程 pool for an idle
Arbortext PE 子进程. The
Arbortext PE 子进程 pool may start a new
Arbortext PE 子进程 if there are fewer than the maximum set by the
maxSubprocesses parameter (see
The minSubprocesses and maxSubprocesses Attributes). However, the scheduler will not wait for a
Arbortext PE 子进程 to become free if the maximum number of
Arbortext PE 子进程 are running and none are idle.
2. When the Queued Transaction Scheduler starts the execution of a queued transaction, it associates the allocated Arbortext PE 子进程 with the transaction, so that any request to allocate a Arbortext PE 子进程 by the queued transaction is satisfied by the transaction that was just allocated. Any attempts to acquire an Arbortext PE 子进程 by a queued transaction are ignored. The allocated Arbortext PE 子进程 is released by the Queued Transaction Scheduler after the queued transaction finishes executing.
An alternate way to allocate an Arbortext PE 子进程 to an immediate request is by using any dynamic component running in the Arbortext PE Request Manager to allocate an Arbortext PE 子进程 by querying the list of configured pools directly. The dynamic component can call the pool's allocate method, which bypasses the check for whether a pool should process a particular request. Use this approach with caution.