The following message is output, and the buffer pool for storing messages runs out of space.

KFCA00854-E insufficient memory in message buffer pool.

• The date and time when the event occurred
• The name of the user server where the event occurred
• Execute the scdls command to check the status of a schedule queue (keep a copy of the execution results)
• Execute the trnls -ta command to check the status of pending transactions (keep a copy of the execution results)

• The information in Table 11-1
• If time passes and the error event continues to occur, execute the command dcsvstop -fd user-server-name to forcibly terminate the user server if possible. If you are able to execute the command, acquire the data in $DCDIR/spool after doing so.

The message KFCA00854-E is output when the size of the buffer pool for storing messages (as specified by the message_store_buflen operand) for the particular user server is inadequate. Common causes of this problem are as follows:

• The value specified for the message_store_buflen operand is too small.
• Too few user servers are permitted to run concurrently (as specified by the parallel_count operand).
• The number of issued RPC requests was greater than expected.
• A service function of the user server is taking a long time to process, causing a backlog of RPC requests in the schedule queue.

Investigate the cause of the error using the flow chart below.

This procedure in this flow chart is designed to be performed immediately after the error event occurs. You may be unable to investigate this type of problem if too much time has passed, though the most likely causes will still be the four listed above. If you are able to identify the cause of the problem from the acquired information, take the appropriate action to resolve it. If you are unable to identify the cause, acquire the data and contact the Support Center.

Figure 11-6 Troubleshooting procedure when message KFCA00854-E is output

Changing the value specified for the parallel_count and message_store_buflen operands requires that you review the values specified for certain other operands. The following table lists the operands whose values must be modified when you increase the value specified for parallel_count. Changes can be made to the balance_count operand without affecting any other operands.

Table 11-9 Operands that require modification when the parallel_count operand value is increased

No.	Definition file name	Operand	Remarks
1	Process service definition (prc)	prc_process_count	--
2	Timer service definition (tim)	tim_watch_count	--
3	Transaction service definition (trn)	trn_tran_process_count	--
4	Lock service definition (lck)	lck_limit_for***	For the four operands (ending with user, dam, tam, and mqa), review the values to suit the processing of the applicable user server)
5	User service definition User service default definition (usrrc) System common definition (betranrc) System service common information definition	max_socket_descriptors	--
6	System environment definition (env)	static_shmpool_size	--
7	System environment definition (env)	dynamic_shmpool_size	Review the value for this operand if you change the value of the dam_tran_process_count operand.
8	DAM service definition (dam)	dam_tran_process_count	Review the value for this operand if a user server whose parallel_count operand you change accesses DAM.
9	TAM service definition (tam)	tam_max_trnnum	Review the value for this operand if a user server whose parallel_count operand you change accesses TAM.
10	RAP-processing listener service definition (name varies)	rap_max_client	Review the value for this operand if a user server whose parallel_count operand you change accesses the RAP-processing listener.

Legend:

--: None

Table 11-10 Operands that require modification when the message_store_buflen operand value is increased

No.	Definition file name	Operand	Remarks
1	System environment definition (env)	static_shmpool_size	For details about estimating shared memory requirements, see the applicable manual or the Release Notes.