[c-nsp] EEM BGP

[Tony Varriale]

Well, did a bunch of testing and I am still stuck.  So here's the basic idea 
and config.

When the peer is actually shut, I log a message to syslog (info simplified 
and anonymized to protect innocent).

event manager applet BGPADJ_SHUT
 event syslog occurs 2 pattern "%BGP-5-ADJCHANGE: neighbor 172.16.10.3 Down" 
period 600
 action 100 cli command "enable"
 action 110 cli command "configure terminal"
 action 120 cli command "router bgp 666"
 action 130 cli command "neighbor 172.16.10.3 shutdown"
 action 140 syslog msg "Neighbor 172.16.10.3 shutdown by EEM"

This works great.  Notice action 140.

To turn the peer back up, I would like to wait 60 seconds (probably 10 
minutes in real world) and look for the "Neighbor 172.16.10.3 shutdown by 
EEM" in the syslog as this will tell me when I need to start my timer.

event manager applet BGPADJ_NOSHUT
 event tag bgpevent1 syslog pattern "%BGP-5-ADJCHANGE: neighbor 172.16.10.3 
Down"
 event tag bgpevent2 syslog pattern "Neighbor 172.16.10.3 shutdown by EEM"
 trigger delay 60
  correlate event bgpevent1 and event bgpevent2
 action 100 cli command "enable"
 action 110 cli command "configure terminal"
 action 120 cli command "router bgp 666"
 action 130 cli command "no neighbor 172.16.10.3 shutdown"
 action 140 syslog msg "Neighbor 172.16.10.3 noshut by EEM"

This is the part that does not work.  For the correlation, I want to either 
look for event 1 and 2 or just 2.  1 and 2 is really just a self check.

The apparent problem is that EEM doesn't look at the messages that it 
injects into syslog.  So, the trigger never happens.  And as verification, I 
tried it with event1 or event2.  While watching debug it picks up on event1.

Any ideas?  Recommendations?

tv

----- Original Message ----- 
From: "Clyde Wildes" <cwildes at progrizon.com>
To: "'Tony Varriale'" <tvarriale at comcast.net>; <cisco-nsp at puck.nether.net>
Sent: Tuesday, December 15, 2009 3:31 PM
Subject: RE: [c-nsp] EEM BGP