Technical data

To change the filter associated with a tracepoint, issue another START command
with a different filter. For example, if you follow the previous START command
with:
CTF> START/FILTER=RX "ROUTING CIRCUIT UNA-0"
only trace records with event code RX (receive) will be collected.
The default value is /NOFILTER, meaning that all trace records generated at the
tracepoint are collected. So, to resume collection of all trace records generated at
the ROUTING CIRCUIT entity, you would issue the following command:
CTF> START "ROUTING CIRCUIT UNA-0"
You can also use the /FILTER qualifier in the ANALYZE command to select
trace records with a specific event code. Note that the /FILTER qualifier in the
ANALYZE command merely restricts the amount of information that is displayed;
it has no effect on the amount of trace information collected from tracepoints.
3.9 Problems Running CTF
There are two major potential problem areas when running CTF:
• Loss of trace records
• Failure to connect to remote systems when remote tracing
3.9.1 Loss of Trace Records
Trace records can be lost when they are generated more quickly than they can
be collected and written to the terminal or to a trace file. CTF will inform you if
you have lost any trace records. If you experience this problem, try the following
corrective actions:
• If you are live tracing, try detached tracing instead. The problem is less likely
to occur with detached tracing.
• Reduce the amount of trace data being generated and/or collected. There are
a number of ways to do this:
Use filtering to reduce the number of trace records collected.
Use the /CAPTURE_SIZE qualifier in the START command to reduce the
size of each trace record.
Do not collect trace records from more than one tracepoint at a time.
• Increase the priority of the trace server process using the /PRIORITY
parameter of the START command.
• If the other actions do not solve the problem, use the /BUFFER_SIZE and
/MAXIMUM_BUFFERS qualifiers in the START command to increase the
size and number of CTF internal buffers.
3.9.2 Failure to Connect to Remote Servers
You may see one of two system messages indicating a failure to connect to a
remote server when remote tracing:
• The message:
%SYSTEM-F-INVLOGIN, LOGIN information invalid at remote node
Using CTF 3–13