ASP.NET for Web Designers

4.4 CQS and System Logger error recovery considerations

CQS is dependent on the availability of the log stream in order to provide shared queue services to its IMS client(s). When CQS connects to the log stream, or when CQS issues a read or write request, the System Logger may return a non-zero return code and reason code, indicating some type of problem.

4.4.1 Non-zero return and reason codes from the System Logger

When the System Logger returns a non-zero return code to a CQS request, the reason code indicates the problem. Messages issued by CQS are described in 4.4.2, "Important messages" on page 130. CQS can take one of the following actions, based on the reason code and whether CQS thinks this is a problem that will soon be resolved:

• Wait for the System Logger to correct the problem and issue an ENF48 signal. When CQS gets the ENF48, it continues processing. No messages are issued by CQS. This occurs under the following conditions. The reason code returned to CQS is in parentheses after the reason text below, in case you want to explore this further.

  • Log stream structure full. Offload in progress (860)

  • Logger CF rebuild in progress (861)

  • Logger structure rebuild in progress (862)

  • Logger structure or CF failed (863)

  • No connectivity to CF (864)

  • Staging data set full (865)

  • System Logger address space initializing (891)

• If the System Logger is not available (890) when CQS tries to connect, or becomes unavailable after CQS has already connected, CQS will issue a message (CQS0350W) and wait for the System Logger to start and send an ENF48 signal.

• If CQS receives a bad return and reason code from an IXGWRITE request, it will issue a message indicating the reason, take a structure checkpoint, and continue processing. The possible causes are:

  • Loss of data in the log stream (405)

  • Possible loss of data in the log stream (407)

• If CQS receives a bad return and reason code from an IXGCONN (connect) request, it will issue a message (CQS0350W) indicating the reason, and then take some action depending on the error as follows:

  • Possible loss of data (407)

    CQS abends with a U0014

  • DS directory full (408) or previous offload error (409)

    CQS takes a structure checkpoint and continues processing

• If CQS gets a bad return code and reason code during IXGBRWSE (read) processing for structure rebuild, CQS aborts the rebuild and issues message CQS0354E. Since this is usually done only for structure recovery, you may have to redefine and reinitialize the structure.

4.4.2 Important messages

There are only a few CQS messages that reference the System Logger. These are briefly described below, but are more fully documented in IMS V8 Messages and Codes Volume 1, GC27-1301 where the messages are associated with System Logger return and reason codes. The CQS message numbers range from CQS0350W through CQS0354E.

CQS0350W reason LOG STREAM logstrname STRUCTURE strname

This message is issued to indicate a variety of error conditions. The reason is based on the return and reason code from the System Logger and is provided as text in the message.

System Logger unavailable

If the System Logger is not available when CQS is initialized, or if it terminates while CQS is running, CQS will issue this message with reason SYSTEM LOGGER NOT AVAILABLE. In this case, CQS cannot do any processing (meaning IMS cannot perform any message queue activity). You should immediately restart the System Logger. CQS will wait for the ENF48 signal from the System Logger, then attempt to connect again.

CQS log connect possible loss of data

CQS abends with either a U0014 (if this condition occurs during CQS initialization) or a U0104 (if it occurs any other time). When this happens, delete and redefine the log stream, restart CQS, and take a structure checkpoint.

CQS log connect previous offload error

CQS takes a structure checkpoint to establish a recovery point, issues a delete call to the System Logger to delete log records no longer needed for recovery, and continues processing.

CQS log connect DS directory full - or - CQS log connect previous offload error

CQS takes a structure checkpoint and issues a delete call to the System Logger to delete log records no longer needed for recovery, and continues processing.

CQS log write loss of data - or - CQS log write possible loss of data

CQS takes a structure checkpoint and continues processing. The structure checkpoint establishes a recovery point for the shared queue structure that will not require the (possibly) lost data.

CQS0352E LOG WRITE ERROR reason

Due to buffer size

CQS tried to write a record to the log stream that exceed the MAXBUFSIZE value in the LOGR policy. CQS continues processing. To correct the problem (that is, to be able to write the larger log records), all CQSs must be shut down to disconnect from the log stream. Then delete the log stream and logger structure definitions in the LOGR policy, and change the LOGR policy with a larger MAXBUFSIZE. Then restart CQS. While this is happening, the message queue structures are not recoverable. Take a structure checkpoint immediately.

CQS0353E CQS variable text

Various messages indicating CQS has started or stopped reading

This message merely indicates that CQS is reading the log stream. This may occur during CQS restart to read the system checkpoint record, or it may occur during message queue structure rebuild (recovery). There will be a STARTED and COMPLETED messages and, if a lot of log records are being read, IN PROGRESS messages.

CQS0354E LOG READ FAILED, LOG RECORD COUNT count

Count is the number of records read before the error

This message indicates CQS log stream read processing failed. If it occurs during CQS initialization, then CQS will abend with a U0018. If it occurs during structure rebuild, then rebuild will terminate. This could mean the message queue structure it was trying to rebuild is unusable and may have to be deleted and reinitialized.