|
|
The key information in each log message is the log class and the location code
number, which uniquely identify the logging message for that entity, and the
Parm. You use these numbers to find the explanation and action in the
tables in this section and other sections of this manual. The actions given
should be performed. Only if there is no clear error reported or the
recommended action fails should system level troubleshooting be performed. When
the system level tests are done, it is advisable to use the configuration files
that have most of the logging classes configured for the console. This saves
having to format the log files for the log messages received during the test.
Internal Errors
Each subsystem has a log class defined for internal errors, as listed in
Table 21-1 "Logging Subsystems and Class
Names". If an internal error ever occurs, a log message is sent to the
console. Record the log message exactly and consult this manual for the
recommended action, which in most cases is to contact HP; refer to
Appendix A "Submitting a CR" of this manual.
Some error messages have an extra parameter whose explanation reads "Parm = BFM
result code." The error codes that are produced by the buffer manager are
listed in Table C-1 "Native Mode NS 3000/iX Buffer
Error Codes" and Table C-2 "Compatibility Mode
Buffer Error Codes". Those tables contain all of the error codes returned
by the NS 3000/iX native mode and compatibility mode buffer manager interfaces.
Informative and Non-Critical Errors
The Network Transport logs all normal or informative events to CLAS0005.
Because of the high number generated, messages for this class are not normally
recorded, even to the disk log file. However, if any of the normal events
encounters an error condition, an additional log message is generated that is
recorded with the log file. This is CLAS0003, non-critical error.
An example of log class 3, non-critical error, is shown in
Figure 21-2 "Non-Critical Transport Error ".
This example shows a non-critical, or Class 3, log message for TCP. The first
part describes the log message, including the event, the entity, the internal
event and the log class. In Figure 21-2
"Non-Critical Transport Error ", the internal event is retransmission which
indicates that there was an attempt to retransmit a TCP packet. The parameter
number indicates the sequence number of the packet that was retransmitted. Also
check the explanation and action listed under the Location Code number which is
61 in the example. The TCP PM section describes the protocol module Location
Code.
Figure 21-2 Non-Critical Transport Error
Statistics Logging Example
An example of log class 6, statistics, is shown in
Figure 21-3 "Logging Statistics Example".
Figure 21-3 "Logging Statistics Example"
shows a statistics log message for TCP. This example shows an informative, or
Class 3, log message for TCP. The first part describes the log message,
including the event, the entity, the internal event and the log class. The
internal event is Statistics, which indicates that the logging message contains
statistical information. For more information check the log file of the remote
node.
Figure 21-3 Logging Statistics Example
The key information in each log message is the log class and the location code
number, which uniquely identify the logging message for that entity, and the
parameter. You use these numbers to find the explanation and action in this
section and other sections of this manual. The "Statistics:" section displays
the following fields:
IPC Port ID. Gives the address of the NetIPC port
associated with the TCP (or PXP) protocol module for this connection.
This is the number to use in matching a TCP/PXP Statistic log message
to a NetIPC log message. Look for the identical number called the
Socket Port ID in the NetIPC log message.
Bytes sent/received. Shows the total number of bytes
transmitted or received by the PM.
User pkts sent/recd. Shows the number of packets sent or
received. May be more than the number of User Sends/Receives because
TCP may need to fragment the message into link-size packets.
PM pkts sent/recd. Shows the number of non-data carrying
packets sent or received, such as window updates and
acknowledgments.
Cn ass pkts sent/recd. Shows the number of connection
assurance packets sent or received. A large number of these indicate
that the connection is idle for long periods of time.
User sends/receives. Shows the number of messages sent or
received.
Pkts discarded/rtx'ed. The number of packets discarded or
retransmitted. If there is a problem with the connection, it may be
reflected here by an extremely high number for one of these fields.
Check for Class 3 logging messages; refer to the specific location
code for more information.
Checksum errors. Shows the number of checksum errors for
this connection. Check for Class 3 logging messages; refer to the
specific location code for more information.
Smooth R/T delay. Shows the observed delay (in milliseconds)
required to receive acknowledgments from the remote node. This number
is used by TCP to dynamically calculate the retransmission
timer.
Init recv/send window. Shows the size of the window used
for flow control. The size varies for each service.
Recv/Send burst size. Shows the burst size for the receiver
and the sender. This value is specified with the IPCCONNECT
used to initiate the connection.
Max send/recv size. Shows the maximum message size for the
receiver and the sender. This value is specified with the
IPCCONNECT used to initiate the connection.
Local/Remote SAP. Shows the TCP Service Access Point
address for the local and the remote node. If you do not know which
application to check on the remote node, use these numbers (reversed)
to find the matching PM in the log file for the remote node. Specify
a time range when formatting the remote node log file to narrow the
search.
Flags/Flags1/Flags2. Internal state flags.
ICMP msgs recd. Total number of ICMP messages that this
connection has received from the IP module.
Local/Remote IP addr. Shows the IP address for the local
and remote node.
Source quenches. Shows the number of times this connection
has been quenched. A connection is quenched when it receives a source
quench ICMP message from IP.
|