* use PULSESECURE-PSG-MIB in layer7::neoteris, also add model() function & remove neoteris fixup from layer7. tests updated.
* will need new mibs release
1. sorts the interfaces as they are processed to make A/B testing easier
2. adds the interface index to the original when a duplicate is seen
3. there may be other de-duplication code in vendor modules, untouched
while it's not yet complete is already is quite functional.
things like vlan, lldp, connected nodes, etc all work
test coverage also included.
features to be added:
link aggregation master/slave (some strange difference with standard 802.3ad i think)
spanning tree will most likely work too
fan/psu status
modules / entity-mib seems to be strangly formatted
netdisco/netdisco#496 has more background on the issue.
somewhere incorrect data is returned for bp_index mappings when running macsuck, but i cannot reproduce this when using snmp::info directly.
forcing i_index to reload works around the issue, performance impact is minimal.
* make fortinet return a useful interface name.
also fixes missing interface issue due to being unable to map the port
'interfaces - ignoring 54 (no port mapping)'
* update oneaccess, regretfully it seems oneos version6 is a step back when it comes to snmp support.
changes:
* start using oneaccess mibs, not much useable data there yet however
* return the device model
* show os version for version 6 too. not many elegeant ways to handle all the version syntaxes used,
so just dump what resembles a version string & be done with it
* updated documentation
tested on:
one420 - os V5.2R1C12
one521 - os 6.1.rc1patch06
one521 - os 6.1.3
some comments in the code as what works & what doesn't.
* typo
* bump required netdisco-mibs version to 4.014 (not released yet) but needed for oneaccess mibs. might as well fail at the start of the build instead of the end
* 2 tests fixed for oneaccess, 1 more to go
* last test fixed
* a few more changes
but mostly need to have a file change so i can trigger a travis build :)
netdisco/snmp-info#295 for issue
tested on c3750x (stacked & single) & nexus 9332.
c3750x didn't show any slave interfaces for it's port channels.
nexus reported data all over the place, even making port channels slave interfaces of other port channels.
with these changes both devices report correct and complete slave interfaces:
tested on c3750x with lacp channels
tested on nexus 9332pq with lacp channels (vpc), static aggregation, vpc peer link, fex uplink.
no regressions detected after a month of testing (we also have a bunch of ciscosb & other ios devices running , but none of those have port channels).
updated documentation, hope it makes sense.
things that didn't work before and still don't:
* mapping vpc portchannels members on both vpc cluster switches (sounds like fun)
* cisco ucs fabric interconnects (nexus 6248 on 3.2.3d) don't show port channel members, but didn't do so before either.
munge_port_ifindex() can most likely be written more elegant, but that's something for a real perl monk.
massive amount of trailing whitespace cleanup
i ran this for 2 weeks, did not notice any regressions.
during those 2 weeks i went over all the diffs on 3 different occasions to make sure i did not delete a comma, quote or whatever.
ran the complete test set as well, no errors there either.
now it's finally commit time then i guess, so when the next module gets based on an existing one we have a clean layout.
for snmp::info device information always comes in handy. add it to the issue template.
basic info which helps: vendor, model & os plus an snmpwalk of the system leaf of snmpv2-mib.
