Le 11/11/15 15:57, Matthew Nickerson a �crit :
Actually, my biggest "problem switch" is an EX8216! That said, it has a
very high port density.
Ah EX8XXX: very good switch chassis; unfortunelaty juniper have made the choice to put small pcc processors on REs.
And with the number of ports, walking the mib can be endless.
I have couple of EX3300 Virtual chassis with eitgh members who suffers the same problems.
We actually cacti have a cacti server ready to use here. And perhaps I
could use the 'thold' plugin to generate syslog message, sending them to
the check_mk event console. This might be the way to go.
I didn't say that using cacti is the right solution (well it's what I use for graphing my devices). I say that the way that cacti poll device is the only good way for polling juniper device like theses.
You can use specific nagios probes also (I you want I can share mine).
Another tips are:
- polling frequencies : 5min should be OK in most case
- snmp can be tweaked a bit on juniper side, (filer interface, filter duplicate, etc...)
Regards,
···
--
Raphael Mazelier