NonStop Operations Guide for H-Series and J-Series RVUs

If you did not have TFDS configured to take the processor dump, you can use the RCVDUMP utility
to perform the dump as described in “Using RCVDUMP to Dump a Processor to Disk” (page 135).
Loading a Down System to Perform a Processor Dump to Disk
To perform a processor dump to disk from a down system, load the system from CPU 0 or CPU 1
with the CIIN file disabled, thereby preventing other processors configured to reload from doing
so and allowing you to dump their memory to disk. For detailed instructions, see “System Load to
a Specific Processor” (page 183) and “Performing a System Load From a Specific Processor”
(page 187).
Using RCVDUMP to Dump a Processor to Disk
Using RCVDUMP to dump a processor to disk involves several preparatory steps, followed by
performing the RCVDUMP command with the proper syntax.
Before You Begin
• You must have a running CPU and a TACL process. The processor in which the TACL process
is running performs the dump.
• If dumpfile already exists, it must be empty. (Its end-of-file pointer, or EOF, must be zero.)
• You must not prime or reset the processor before performing a processor dump.
To prepare for a disk dump:
1. Verify that a disk is available with enough space to store the dump.
2. To empty an existing dumpfile:
> FUP PURGEDATA dumpfile
Dumping a Processor Using RCVDUMP (Excludes NonStop NS16000 Series and NS14000 Series
Systems)
NOTE: See “Dumping a Processor Using RCVDUMP for NonStop NS16000 Series and NS14000
Series Systems” (page 136) for the procedure for NS16000 series and NS14000 series systems.
To use RCVDUMP to dump a processor on a running system:
1. Log on to a TACL session as the super ID (255,255).
2. At a TACL prompt, run the RCVDUMP utility, choosing the parameters and options appropriate
for your scenario:
NOTE: If you include the SLICE or the PARALLEL parameter in a RCVDUMP command to
any NonStop system other than a NonStop NS16000 series or NS14000 series system, a
RCVDUMP error is generated to report: parameter is not supported on Non-NSAA.
RCVDUMP filename, cpuNum [, SLICE bladeId]
[, START startAddress][, END endAddress]
[, ONLINE | PARALLEL][, FULL]
filename
is the name of the disk file to which the dump is to be written.
cpuNum
is the number of the logical processor from which a processor element is to be dumped.
Specify cpuNum as an integer in the range from 0 through 15.
Recovery Operations for Processors 135