Viewing a System Trace: Diagnostics Example

This subject is covered in the help topic To view a system trace. However, the options are complex and it is worth looking at an example.

To obtain a trace of the system trace table:

  1. In the Enterprise Server Administration screen, click the diagnostic server's Details button to access the server's Details screen.
  2. On the Details screen, click ES Monitor & Control to access the diagnostic server's ESMAC screen.
  3. In the Diagnostics area in the left pane, check Trace, and select either the A or B to select the A or B dataset.
    Note: Do not select C, since the in-memory trace table belongs to the diagnostics server.
  4. In the Blocks field, specify the number of trace blocks that are to be aggregated into one trace index entry.
  5. Click Display.
Figure 1. Trace Index Page
Trace Index Page

Each block is the same size as the system trace table, that is, it contains the number of trace entries specified in Trace Table Size on the Edit Server page for the server. The index entries are displayed with a timestamp against each one. This is the timestamp of the first trace entry in each set of blocks. These timestamps help you narrow down your search to particular events.

To view the actual trace, go to one of the index entries, and make selections in the following fields:

The Level, PIDs, and Task IDs controls are filters that help you select just the specific information of interest. If you leave the level as 0, the default, and do not check any process IDs or task IDs, you will see the minimum level of information for all processes and tasks that have entries in the block represented by the trace index entry.
Note: It is possible to make selections such that no trace information is generated. This happens if you select both processes and tasks, and none of the selected tasks ran on the selected processes; a task cannot run on more than one process. If a task requires communications work, that work has its own process and task, and there is no automated way of relating the two task IDs. You always see the first trace entry in the block, irrespective of filtering.

The following is an example of how you can use the controls to select the information you want to see. Suppose you specify 10 in Blocks in the Diagnostics group of the menu. There are actually 60 blocks of information in the dataset, so 6 index entries are displayed. Suppose you were interested in a particular SEP, which you can see listed in the third and fourth index entries. You might then specify 20 in Blocks for the third index entry, check the box next to the process ID for the SEP, then click Details. Now you can only see trace entries for that SEP, in all the blocks that contain them, starting with the first block of the third index entry.

Each trace entry contains the following information:

An interpretation of the traced event, if one can be supplied.
Sequence number of the entry in ascending order.
The five-digit task number.
The five-digit process ID.
The ID of the event, command, or instruction that was traced. This is a four-byte hexadecimal number.
A time stamp that indicates when the trace entry was written. The time stamp is in hours, minutes, seconds, and hundredths of a second.
aaaa bbbb
Eight bytes of entry-specific data

When you have finished viewing a trace, click Back to return to the trace index page.