TRANSFER Installation and Management Guide

Using the Logs for Troubleshooting
Troubleshooting
13198 Tandem Computers Incorporated 12–21
Using the Logs for
Troubleshooting
You most commonly use the asynchronous logs for solving problems in two ways:
To determine what happened at a particular time
Search the logs for the date and time. Limit your first search to the hour and
minute. If you specify the second, you might not find an entry. If you specify only
the hour, you might get a large area to go through; on a busy system, an hour can
span many spooler pages of a log.
After you locate the time in the log, examine the log to see if it tells you what went
wrong.
To determine what happened to a particular item
To trace the progress of a specific item, search for the item ID in each log. After
you find all the occurrences, correlate the time and the thread numbers to track the
item.
You can most efficiently use the asynchronous logs for tracking problems by enabling
all three logs on a single system and specifying the log files to nonprinter spooler
locations rather than to a disk file or a terminal.
You can most easily change the log locations on the TMANAGER LOG CONTROL
screen, discussed in Section 10 of this guide. You can also modify the locations using
the TRANSFER configuration program, CDTRDEF.
A spooler job provides the following advantages:
A history of events. A job stays until you print or delete it.
A search mechanism. With the FIND command in PERUSE, you can quickly
locate an item by its item ID or the time.
An easy mechanism for printing either the entire log or only the relevant parts of a
log. With the LIST command in PERUSE, you can specify the parts you want:
LIST /OUT file name/ page/page C
The C at the end of the page range gives you the controls and setmodes in the
original data; otherwise, they are ignored.
A way of separating TWORK and TRECV processes. If there is more than one
process for TWORK or TRECV, each process has its own spooler job. With a
terminal or a disk file, the entries are mixed.
A mechanism for deleting unnecessary jobs. When the LOGRECSPEROPEN limit
is reached, the job is closed and a new one is created.