The SDRs are stored
together in a self-relative set. This self-relative set is called a Support
Record Collection. Each individual SDR is identified with a record-id. The
record-id of the most recent SDR is always 0 (zero). The next older SDR is
record-id 1, and so on, for the number of records in the stored collection. For
example, if there are five SDRs, they are identified as SDR-0 through SDR-4.
Figure 2. Support Data Collection Hierarchy
When a new SDR is
created, the numbers all increment by one and the newest SDR is given the value
of 0. If the total number of records exceeds a configured maximum, then the
oldest SDR is deleted.
Using the example
above, when the maximum SDR count of 5 is reached, the SDRs continue to be
SDR-0 through SDR-4, with the file timestamps indicating that the files are
changing over time.
The time interval
between collections may vary by several minutes in relation to the specified
sleep-duration. This is because the interval specifies the idle time between
scheduled collection runs. Since the actual overhead of the collecting process
is not included in the scheduled intervals, the time differences between
collections includes this non-deterministic amount of time.
Important:
Using a shorter
interval to compensate for this behavior is
not
recommended, since it will only add to the overhead incurred by the collection
process and will ultimately impact the overall system performance. The
sleep-duration (idle-time) between scheduled collections is an important
component of the "self-throttling" mechanism that should not be circumvented by
the user.
The Exec Mode
show support
collection command displays useful information about the Support Data
Collector. The output includes information about when the collector last ran,
how long it took to run, when it is scheduled to run again, as well as the
number of SDRs currently stored, where they are stored, and how much storage
space is being used. Refer to
Exec Mode Commands for
more detail about this command.