> From: Harrington, David [mailto:[EMAIL PROTECTED] > Rather than reinventing wheels, let's try to review work that has > already been done in the IETF for these purposes, and discuss any > experience that has already been had with these approaches. > Please look > at the RFC3165 Script MIB > (ftp://ftp.rfc-editor.org/in-notes/rfc3165.txt) and the NETCONF work > (http://www.ietf.org/html.charters/netconf-charter.html).
That sounds like the thing to do for me. FC31655 MIB sounds much like the proposed "B2" option. I didn't have time to look at netconf in particular, but probably this is a better route to take. I think the overall idea behind a standard MIB is the ability to configure important base parameters in a vendor-ignorant way. If we go ahead and upload config files, this is not vendor-irgnorant. So I doubt that this is actually that useful in regard to what the original idea was (I may be wrong here). This brings me to the point of questioning if it really makes sense to try to include that functionality in syslog-mib. Maybe it would be better to take it off (simplyfying things), look if netconf or 3165 works and after some time-revaluate syslog-mib usage? I have the impression that can save us a lot of time without loosing much. But again, I have to say I am not very deep inside SNMP issues ;) > > The Script MIB is all about uploading/downloading scripts to > execute on > a managed device. It is language-independent, but recognizes the > likelihood of Java, Perl, and CLI scripts. > This MIB has an (optional) on-line editing capability, which > experience > has shown few operators want to use; they would rather get the script, > modify it offline, and send it back to the device. generally speaking, my own experience is the same. Operators love to edit files which whatever prefrerred editor they have. Hard to believe the will use another one. The only important feature is an import/export of all the config file lines. > > The NETCONF work is trying to make it possible to > download/upload script > files for configuring devices, as a way to standardize CLI scripting. That sounds useful. Rainer
