On Wed, 26 Sep 2001 10:00:58 -0700, Donna Garverick <[log in to unmask]>
wrote:

>Dave Anderson wrote:
>
>> Well Donna I may need a beating to get this thru my thick skull.
>
>(i rather doubt that :-)
>
>> I set up syslog but it is not logging anything.  Well it does log the
>> startup but that is all.
>
>there's very little mpe software that currently logs to syslog -- mostly
>because we've always had such good logging built into the o.s.

Although the format is a little cryptic, you CAN search the MPE logs "in
real time" much like you would be scanning a syslog file.  While not
promoted very much, MPEX with VEAUDIT does have the ability to search MPE
log files for LOGON, CONSOLE, and FCLOSE records -- take a look at the "%
VEAUDIT LISTLOG" commands  (err, was that a <plug>?  I haven't worked for
them for over 5 years now...)

One thing to watch for (especially for FCLOSE records) is the fact that the
log record will only contain the JSID number -- the "details" of who that
is (or was) have to be extracted from a corresponding "job initiation" log
record.  (in other words, FLCOSE would report that session #s1234 "closed"
the very important file you are looking for with "purge" disposition,
however you don't know from this whether #s1234 is MANAGER.SYS or
SNEAKY.HACKER...)

>however -- these days when so many of us are running mpe systems in mixed-
o.s.
>shops, our  inability to (easily) tell the 'world' what's happening on our
mpe
>boxes is (imo) a 'black eye'.  while i realize that there are $olutions
>available (for example, openview), i don't consider them to be the 'final
>answer' :-)

Have the "standard" (?) C-library calls for "syslog" been ported?
Obviously, they have at some level because syslog does exist and runs on
MPE, but the real question is are these supplied in a library from HP
(meaning HP's C compiler can call them) or are they "freeware" (called via
gcc...)?  If they are indeed available & supported straight from HP, I'd
suggest an "interim" solution of a home-brew mpe-to-syslogd program that
scans for "job init" records and posts a message to syslog -- this in turn
would allow for a more "focused" program to view/report/act-upon repeated
failures...

Tom

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *