pmnsmerge(1) performs the same function as pmnsadd and is faster, more robust and more flexible. It is
therefore recommended that pmnsmerge(1) be used instead.
pmnsadd adds subtree(s) of new names into a Performance Metrics Name Space (PMNS), as used by the
components of the Performance Co-Pilot (PCP).
Normally pmnsadd operates on the default Performance Metrics Name Space (PMNS), however if the -n option
is specified an alternative namespace is used from the file namespace.
The default PMNS is found in the file $PCP_VAR_DIR/pmns/root unless the environment variable PMNS_DEFAULT
is set, in which case the value is assumed to be the pathname to the file containing the default PMNS.
The new names are specified in the file, arguments and conform to the syntax for PMNS specifications, see
PMNS(5). There is one PMNS subtree in each file, and the base PMNS pathname to the inserted subtree is
identified by the first group named in each file, e.g. if the specifications begin
myagent.foo.stuff {
mumble 123:45:1
fumble 123:45:2
}
then the new names will be added into the PMNS at the non-leaf position identified by myagent.foo.stuff,
and following all other names with the prefix myagent.foo.
The new names must be contained within a single subtree of the namespace. If disjoint subtrees need to
be added, these must be packaged into separate files and pmnsadd used on each, one at a time.
All of the files defining the PMNS must be located within the directory that contains the root of the
PMNS, this would typically be $PCP_VAR_DIR/pmns for the default PMNS, and this would typically imply
running pmnsadd as root.
As a special case, if file contains a line that begins root { then it is assumed to be a complete PMNS
that needs to be merged, so none of the subtree extraction and rewriting is performed and file is handed
directly to pmnsmerge(1).
Provided some initial integrity checks are satisfied, pmnsadd will update the PMNS using pmnsmerge(1) -
if this fails for any reason, the original namespace remains unchanged.