pmlogger automagically adds pmcd.pmlogger.{host,port,archive} self-descriptions
to its output archives, though this is not documented in the pmlogger man page
and represents potentially undesirable information leakage. (They are briefly
mentioned in pmlogextract.1.) The only use of these metrics that a quick
git-grep identifies is in the pmsnap program, which uses the
pmcd.pmlogger.{host,port} fields to issue a 'pmlc flush'.
However, pmlc flushing has become a no-op since pmlogger started issuing
semantic writes (pcp 3.9.3ish in 2014). Moreover, the pmcd.pmlogger.host name
has the usual FQDN risks (may have some sort host name in it, but there's no
guarantee that it is a remotely- (or even locally-) tcp/ip resolvable one).
If all this obsoleteish functionality is to be retained, please consider
documenting it in pmlogger.1.