TFDS Manual

Using TFDS
Tandem Failure Data System (TFDS) Manual—520628-003
2-4
Monitoring TFDS Status
•
If no DUMPVOLUME was specified or there was insufficient space on the volume,
TFDS searched all allowed volumes for the volume with the most available space.
The dump files and incident-related data were stored there.
TFDS operation in FMD mode changes only slightly from this model. The
ALTERNATE-VOLUMES command allows you to specify secondary volumes in
addition to the primary volume specified in the DUMPVOLUME command. Alternate
volumes are used (if space is available) before TFDS searches all volumes for the one
with the most space available.
For HP NonStop S76000 and S86000 servers with 4 or 16 gigabytes of memory, at
least 20 gigabytes of disk space should be free for use by TFDS on the primary volume
(or alternate volumes). For a NonStop S76000 or S86000 with 16-gigabytes of
memory, all 20 gigabytes must be free on the same volume. TFDS does not spread
files from a single incident across volumes or subvolumes. 20 gigabytes is enough
space to process a single halted processor with 16-gigabyte memory. Unless additional
disk resources are available across the system, move the incident data to tape as
quickly as possible (see the BACKUP command on page 4-10) to free the disk space.
Monitoring TFDS Status
To check the status of TFDS incident processing, use one of the following methods:
•
To view TFDS ($ZDMP) events, use an event viewer such as WebViewPoint or the
TSM EMS Event Viewer Application. For a description of TFDS events, see
Appendix D, EMS Messages and Templates.
•
To view processor status and reload states, use the TFDSCOM STATUS command
described on page 4-36.
Caution. Although TFDS can accept several dump requests simultaneously, several
RCVDUMPs running concurrently place an extreme load on a system. For a description of
using this configuration parameter to limit the number of concurrent dumps TFDS can run, see
MAXCONDUMPS
on page 4-29.
Note. To obtain all pertinent events related to TFDS, filter the DMP, CPU, and NSC
subsystems simultaneously.