User's Guide

Troubleshooting HyperFabric
Running Diagnostics
Chapter 6
118
packet_data or 0x0004 Trace the packet data.
packet_hdr or 0x0008 Trace the protocol headers (TCP/UDP/IP).
data_struct or 0x0010 Trace the HyperFabric software CLIC-specific
data structures and events.
rcv_path or 0x0020 Trace the HyperFabric software receive path.
control or 0x0040 Trace the HyperFabric software control
messages.
all or 0xffff Turn on all possible trace levels. Note that this
will have a severe impact on performance.
To find out what the current tracing level is, run this parameter (-T) without any
keyword or hexadecimal value specified.
-D specifies that you want to dump the trace buffers from the kernel to a user-space
file and format the data using the formatter TCP_IP. The HyperFabric software trace
data is dumped into the file /var/adm/clic_ip_drv.trc and formatted.
-C TCP_IP operates in a way similar to -D TCP_IP, except that the trace data in the
kernel is dumped at regular intervals to the trace file /var/adm/clic_ip_drv.trc0,
and if that file gets full, to /var/adm/clic_ip_drv.trc1. So, new trace events are
appended to the trace output file. This stops when tracing is turned off.
-B specifies that you want the size (in bytes) of the trace buffer to be
trace_buffer_size
. The buffer is dynamically allocated when tracing is enabled.
Also, the buffer is circular, which means that when the end of the buffer is reached,
the data wraps around to the beginning of the buffer (and overwrites any previous
data). If you do not specify this parameter, a default buffer is created with a size of
64k bytes.
-F specifies
trace_file_size
as the maximum size of the trace output file. For the
-D TCP_IP parameter, the file is /var/adm/clic_ip_drv.trc. For the -C TCP_IP
parameter, the files are /var/adm/clic_ip_drv.trc0 (and if needed,
/var/adm/clic_ip_drv.trc1).
-d dumps the memory of the adapter identified by
adapter_ID
to the dump file
filename
. If you do not specify
filename
, the default file is
/var/adm/clic_fw.dump
x
, where
x
is the adapter instance number.
-? displays the online help for clic_diag.
If you do not specify any of the above parameters, the online help for clic_diag is
displayed.
Example of clic_diag
An example of the clic_diag command is shown below.
If the local node is bently6, and you want to confirm that all of the adapters on bently6
are communicating with the target adapters on bently8, issue this command:
clic_diag -r bently8
The generated output could look like this:
CLIC_PROBE: 256 byte packets
Source adapter id: bently6.corp4.com:clic0
Target adapter id: bently8.corp2.com:clic1