On Tue, 15 Nov 2011 10:58:20 -0600 Greg Cooper gwc2004@gmail.com wrote:
Yes, it was the bridge plugin. It seems to time out in about 4 seconds... Here's some debug output:
[snip]
2011-11-15 17:52:12,867 [DEBUG jobs.jobhandler] [inventory sw01-3750.] Plugin nav.ipdevpoll.plugins.bridge.Bridge(u'sw01-3750.') reported a timeout
[snip]
SNMP is configured properly on the 3750. What do you think?
The Bridge plugin tries to figure out which interfaces on a device are switch ports. Cisco complicates matters by using alternate BRIDGE-MIB instances per. VLAN, making community indexing necessary.
I suspect the plugin sometimes might query with a modified community that the switch doesn't respond to, causing a timeout. Unfortunately, that derails the entire collection job. I would consider making the plugin ignore timeouts and continue to the next query.
On the other hand, your 3750 may just be plain slow at responding to SNMP queries, as indicated by one of your other posts. Unfortunately, ipdevpoll has no mechanism to adjust timeout values individually.
There may be two locations in the code where you could patch NAV to increase the timeout on snmpwalks for all devices; if you care to try, I can write a quick patch.