> Perhaps I didn't look close enough, but all of those > require yet > another agent be installed on the system, as well as > a rather large > amount of agent configuration to get it to where it > can actually get > anything more than the bare minimum of information > from an Opensolaris > system.
I had three problems with the scripts I wrote where I used to work: 1) They originally ran at the same time agents from Tivoli ran. Not only did the Tivoli agents pound on the systems, but they provided 0 value. Badly written agents affect metrics gathering in exactly the way they are not supposed to--they directly measure little but their own impact on the systems. 2) The scripts originally all ran at the same time, stomping on the metrics-gathering system. 3) Even when the above two issues were "fixed" by changing the times and putting in a random-wait clause, their impact remained well above non-zero. I agree with Jason; a decent Net-SNMP module that can feed useful, robust data into whatever pre-existing SNMP framework our "clients" may have seems a logical, least impact way forward. With your (Adrian) knowledge of good, useful (Open)Solaris metrics, I would love to see your input into what the MIB should measure and how. While it's impossible to be perfect right out of the box, the better the MIB starts out, the more adoption (and maybe continued development) it should see. Rainer -- This message posted from opensolaris.org _______________________________________________ sysadmin-discuss mailing list sysadmin-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/sysadmin-discuss
