openafs/doc/man-pages/pod8/fstrace_dump.pod

=head1 NAME

fstrace_dump - Dumps a trace log

=head1 SYNOPSIS

=for html
<div class="synopsis">

B<fstrace dump> S<<< [B<-set> <I<set name>>+] >>>
    S<<< [B<-follow> <I<log name>>] >>>
    S<<< [B<-file> <I<output filename>>] >>>
    S<<< [B<-sleep> <I<seconds between reads>>] >>>
    S<<< [B<-debug>] >>>
    [B<-help>]

B<fstrace d> S<<< [B<-se> <I<set name>>+] >>>
    S<<< [B<-fo> <I<log name>>] >>>
    S<<< [B<-fi> <I<output filename>>] >>>
    S<<< [B<-sl> <I<seconds between reads>>] >>>
    S<<< [B<-d>] >>>
    [B<-h>]

=for html
</div>

=head1 DESCRIPTION

The B<fstrace dump> command displays the current contents of the C<cmfx>
trace log on the standard output stream or writes it to the file named by
the B<-file> argument.

To write the log continuously to the standard output stream or to a file,
use the B<-follow> argument. By default, the log's contents are written
out every ten seconds and then automatically cleared. To change the
interval between writes, use the B<-sleep> argument.  To display the raw
format of each trace log entry, use the B<-debug> argument.

=head1 CAUTIONS

This command produces output only if the C<cm> event set is active. To
display or set the event set's state, use the B<fstrace lsset> or
B<fstrace setset> command respectively.

To make the output from this command maximally readable, the message
catalog file called F<afszcm.cat> must reside in the local
F</usr/vice/etc/C> directory. If necessary, copy the file to that
directory from the AFS Binary Distribution before activating tracing.

When the C<cm> event set is active, a defined amount of kernel memory (by
default, 60 KB) is allocated for the C<cmfx> trace log. As described in
L<fstrace(8)>, when the buffer is full, messages are overwritten in a
circular fashion (new messages overwrite the oldest ones). To allocate
more kernel memory for the log, use the B<fstrace setlog> command; to
display the log buffer's current size, use the B<fstrace lslog> command
with the B<-long> argument.

=head1 OPTIONS

=over 4

=item B<-set> <I<set name>>+

Names the event set for which to write out the associated trace log. The
only acceptable value is C<cm> (for which the associated trace log is
C<cmfx>). Provide either this argument or the B<-log> argument, or omit
both to write out the C<cmfx> log by default.

=item B<-follow> <I<log name>>

Names the trace log to write out continuously at a specified interval (by
default, every ten seconds; use the B<-sleep> argument to change the
interval). The log is cleared after each write operation.

The only acceptable value is C<cmfx>. Provide either this argument or the
B<-set> argument, or omit both to write out the C<cmfx> log by default.

=item B<-file> <I<output filename>>

Specifies the pathname of the file to which to write the trace log's
contents. It can be in AFS or on the local disk. Partial pathnames are
interpreted relative to the current working directory. If this argument is
omitted, the trace log appears on the standard output stream.

=item B<-sleep> <I<seconds between reads>>

Sets the number of seconds between writes of the trace log's contents when
it is dumped continuously. Provide the B<-follow> argument along with this
one. If this argument is omitted, the default interval is ten seconds.

=item B<-debug>

Specifies that each record in the trace log should be displayed in its raw
hex format as well as the normal decoded output.

=item B<-help>

Prints the online help for this command. All other valid options are
ignored.

=back

=head1 OUTPUT

The output begins with a header specifying the date and time at which the
write operation began. If the B<-follow> argument is not included, the
header also reports the number of logs being dumped; it is always C<1>,
since there is only the C<cmfx> trace log.  The format of the header is as
follows:

   AFS Trace Dump -
     Date: I<starting_timestamp>
   Found 1 logs.
   Contents of log cmfx:

Each subsequent message describes a Cache Manager operation in the
following format:

   time <timestamp>, pid <pid>:<event_message>

where

=over 4

=item <timestamp>

Specifies the time at which the Cache Manager performed the operation, as
the number of seconds since the dump began.

=item <pid>

Specifies the process ID of the process or thread associated with the
message.

=item <event_message>

Is the message itself. They are generally meaningful only to someone
familiar with the AFS source code.

=back

In addition, every 1024 seconds the fstrace command interpreter writes a
message that records the current clock time, in the following format:

   time <timestamp>, pid <pid>: Current time: <unix_time>

where

=over 4

=item <timestamp>

Is the number of seconds from the start of trace logging.

=item <pid>

Is the process ID number.

=item <unix_time>

Is the machine's clock time, represent in the standard UNIX time format as
the number of seconds since midnight on January 1, 1970.

=back

Use this message to determine the actual clock time associated with each
log message. Determine the actual time as follows:

=over 4

=item *

Locate the message of interest.

=item *

Search backward through the trace file for the closest current time
message.

=item *

If the current time message's timestamp is smaller than the log message's
timestamp, subtract former from the latter.  If the current time message's
timestamp is larger than the log message's timestamp, add 1024 to the
latter and subtract the former from the result.

=item *

Add the resulting number to the current time message's <unix_time> to
determine the log message's actual time.

If any of the data in the kernel trace buffer has been overwritten since
tracing was activated, the following message appears at the appropriate
place in the output:

   Log wrapped; data missing.

To reduce the likelihood of overwriting, use the B<fstrace setlog> command
to increase the kernel buffer's size. To display the current defined
buffer size, use the B<fstrace lslog> command with the B<-long> argument.

The following message at the end of the log dump indicates that it is
completed:

   AFS Trace Dump - Completed

=back

=head1 EXAMPLES

The following command dumps the log associated with the cm event set to
the standard output stream.

   # fstrace dump -set cm
   AFS Trace Dump -
      Date: Tue Apr  7 10:54:57 1998
   Found 1 logs.
   time 32.965783, pid 0: Tue Apr  7 10:45:52 1998
   time 32.965783, pid 33657: Close 0x5c39ed8 flags 0x20
   time 32.965897, pid 33657: Gn_close vp 0x5c39ed8 flags 0x20 (returns 0x0)
   time 35.159854, pid 10891: Breaking callback for 5bd95e4 states 1024 (volume 0)
   time 35.407081, pid 10891: Breaking callback for 5c0fadc states 1024 (volume 0)
                                    .
                                    .
                                    .
   time 71.440456, pid 33658: Lookup adp 0x5bbdcf0 name g3oCKs \
        fid (756 4fb7e:588d240.2ff978a8.6)
   time 71.440569, pid 33658: Returning code 2 from 19
   time 71.440619, pid 33658: Gn_lookup vp 0x5bbdcf0 name g3oCKs (returns 0x2)
   time 71.464989, pid 38267: Gn_open vp 0x5bbd000 flags 0x0 (returns 0x0)
   AFS Trace Dump - Completed

The following command dumps the trace log associated with the cm event set
on the local machine to the file C<cmfx.dump.file.1>, using the default
interval of 10 seconds between successive dumps:

   # fstrace dump -follow cmfx -file cmfx.dump.file.1

=head1 PRIVILEGE REQUIRED

The issuer must be logged in as the local superuser C<root>.

=head1 SEE ALSO

L<afszcm.cat(5)>,
L<fstrace(8)>,
L<fstrace_lslog(8)>,
L<fstrace_setlog(8)>,
L<fstrace_lsset(8)>

=head1 COPYRIGHT

IBM Corporation 2000. <http://www.ibm.com/> All Rights Reserved.

This documentation is covered by the IBM Public License Version 1.0.  It was
converted from HTML to POD by software written by Chas Williams and Russ
Allbery, based on work by Alf Wachsmann and Elizabeth Cassell.