Every request to administer an WebSphere MQ Everyplace resource takes the same basic form. Figure 13 shows the basic structure for all administration request messages:
A request is made up of:
Figure 13. Administration request message
The base administration fields, that are common to all administration messages, are:
Table 3. Administration actions
Administration action | Purpose |
---|---|
MQE_ADMIN_ACTION_CREATE | Create a new instance of a managed resource. |
MQE_ADMIN_ACTION_DELETE | Delete an existing managed resource |
MQE_ADMIN_ACTION_INQUIRE | Inquire on one or more characteristics of a managed resource |
MQE_ADMIN_ACTION_INQUIREALL | Inquire on all characteristics of a managed resource |
MQE_ADMIN_ACTION_UPDATE | Update one or more characteristics of a managed resource |
All resources do not necessarily implement these actions. For instance, it is not possible to create a queue manager using an administration message. Specific administration messages can extend the base set to provide additional actions that are specific to a resource.
Each common action provides a function that sets the
MQE_ADMIN_ACTION field:
Table 4. Setting the administration action field
Administration action | Setting function |
---|---|
MQE_ADMIN_ACTION_CREATE | mqeAdminMsg_create(parameters) |
MQE_ADMIN_ACTION_DELETE | mqeAdminMsg_delete(parameters) |
MQE_ADMIN_ACTION_INQUIRE | mqeAdminMsg_inquire(parameters) |
MQE_ADMIN_ACTION_INQUIREALL | mqeAdminMsg_inquireAll(parameters) |
MQE_ADMIN_ACTION_UPDATE | mqeAdminMsg_update(parameters) |
MQeAdminMsgHndl hMsgObj, MQeExceptBlock *pErrStruct, MQECONST MQeFieldsHndl hParms
This field determines how many times an action can be retried if the initial action fails. The retry occurs either the next time that the queue manager restarts or at the next interval set on the administration queue.
A set of functions are available for setting some of the request
fields:
Table 5. Setting administration request fields
Administration action | Field type | Set and get functions |
---|---|---|
MQE_ADMIN_PARAMS | MQeFields | mqeAdminMsg_getInputFields( MQeAdminMsgHndl hMsgObj , MQeExceptBlock * pErrStruct , MQeFieldsHndl * phFields) |
MQE_ADMIN_ACTION | int | mqeAdminMsg_setAction( MQeAdminMsgHndl hMsgObj, MQeExceptBlock * pErrStruct, MQEINT32 action) |
MQE_ADMIN_TARGET_QMGR | ascii | mqeAdminMsg_setTargetQMgr( MQeAdminMsgHndl hMsgObj, MQeExceptBlock * pErrStruct, MQECONST MQeStringHndl hName) |
MQE_ADMIN_MAXATTEMPTS | int | mqeAdminMsg_setMaxAttempts( MQeAdminMsgHndl hMsgObj, MQeExceptBlock * pErrStruct, MQEINT32 maxAttempts) |
Every resource has a set of unique characteristics. Each characteristic has a name, type and value, and the name of each is defined by a constant in the administration message. The name of the resource is a characteristic that is common to all managed resources. The name of the resource is held in the MQE_ADMIN_NAME, and it has a type of ascii.
The full set of characteristics of a resource can be determined by using the mqeAdminMsg_characteristics() function against an instance of an administration message. This function returns an MQeFields object that contains one field for each characteristic. MQeFields functions can be used for enumerating over the set of characteristics to obtain the name, type and default value of each characteristic.
The action requested determines the set of characteristics that can be passed to the action. In all cases, at least the name of the resource, MQE_ADMIN_NAME, must be passed. In the case of MQE_ADMIN_ACTION_INQUIRE this is the only parameter that is required.
By default, no reply is generated, when an administration request is processed. If a reply is required, then the request message must be setup to ask for a reply message. The following fields are defined in the MQe class and are used to request a reply.
If MQE_MSG_STYLE is set to MQE_MSG_STYLE_REQUEST (a reply is required) then the location that the reply is to be sent to must be set into the request message. The two fields used to set the location are:
If the reply-to queue manager is not the queue manager that processes the request then the queue manager that processes the request must have a connection defined to the reply-to queue manager.
For an administration request message to be correlated to its reply message the request message needs to contain fields that uniquely identify the request, and that can then be copied into the reply message. WebSphere MQ Everyplace provides two fields that can be used for this purpose:
Any other fields can be used but these two have the added benefit that they are used by the queue manager to optimize searching of queues and message retrieval. The following code fragment provides an example of how to prime a request message:
MQERETURN primeAdminMsg(MQeQueueAdminMsgHndl hMsg, MQeExceptBlock *pErrorBlock, MQeFieldsHndl * phMsgTest, MQeStringHndl hQueueManagerName) { MQERETURN rc; /* Set the target queue manager that will process this message */ rc = mqeAdminMsg_setTargetQMgr((MQeAdminMsgHndl)hMsg, pErrorBlock, hQueueManagerName); if (MQERETURN_OK == rc) { /* Ask for a reply message */ rc = mqeFields_putInt32((MQeFieldsHndl)hMsg, pErrorBlock, MQE_MSG_STYLE, MQE_MSG_STYLE_REQUEST); if (MQERETURN_OK == rc) { rc = mqeFields_putAscii((MQeFieldsHndl)hMsg, pErrorBlock, MQE_MSG_REPLYTO_Q, MQE_ADMIN_REPLY_QUEUE_NAME); if (MQERETURN_OK == rc) { rc = mqeFields_putAscii((MQeFieldsHndl)hMsg, pErrorBlock, MQE_MSG_REPLYTO_QMGR, hQueueManagerName); if (MQERETURN_OK == rc) { rc = mqeFields_new(pErrorBlock, phMsgTest); if (MQERETURN_OK == rc) { MQEINT64 v; MQeStringHndl hFieldName; /* create some identical data fields in both hMsg and *phMsgTest so that the replay message can be matched against when it is returned to the replyToQ. */ rc = mqe_uniqueValue(pErrorBlock, &v); if (MQERETURN_OK == rc) { rc = mqeString_newChar8(pErrorBlock, &hFieldName, "IDField" ); if (MQERETURN_OK == rc) { rc = mqeFields_putInt64(*phMsgTest, pErrorBlock, hFieldName, v); if (MQERETURN_OK == rc) { rc = mqeFields_putInt64((MQeFieldsHndl)hMsg, pErrorBlock, hFieldName, v); if (MQERETURN_OK != rc) { displayError("mqeFields_putInt64 error (in primeAdminMsg, hMsg)", pErrorBlock); } } else { displayError("mqeFields_putInt64 error (in primeAdminMsg, phMsgTest)", pErrorBlock); } (void)mqeString_free(hFieldName, NULL); } else { displayError("mqeString_newChar8 error (in primeAdminMsg)", pErrorBlock); } } else { displayError("mqe_uniqueValue error (in primeAdminMsg)", pErrorBlock); } } else { displayError("mqeFields_new error (in primeAdminMsg)", pErrorBlock); } } else { displayError("mqeFields_putAscii (2) error (in primeAdminMsg)", pErrorBlock); } } else { displayError("mqeFields_putAscii (1) error (in primeAdminMsg)", pErrorBlock); } } else { displayError("mqeAdminMsg_Int32 error (in primeAdminMsg)", pErrorBlock); } } else { displayError("mqeAdminMsg_setTargetQMgr error (in primeAdminMsg)", pErrorBlock); } return rc; }
When the administration request message has been created, it is sent to the target queue manager using standard WebSphere MQ Everyplace message processing APIs. Depending on how the destination administration queue is defined, delivery of the message can be either synchronous or asynchronous.
Standard WebSphere MQ Everyplace message processing APIs are also used to wait for a reply, or notification of a reply. There is a time lag between sending the request and receiving the reply message. The time lag may be small if the request is being processed locally or may be long if both the request and reply messages are delivered asynchronously. Administration example Ex1 contains code that demonstrates these functions.