RE: help setting up SNMP traps to Xymon

[Nicole Beck <nskyrca (at) syr.edu>]

Hi Andy,
I think you may have answered part of my question. So I have to correctly define traps in my snmptt.conf.oracle file.  The only thing in this file for the trap that's being sent is below.   Given this definition, any trap I get, even if the severity is Critical, is still going to be "Normal" based on this file, thus Xymon won't ever change the status to red.  (For tests, I changed the "Normal" in this file to "Severe", and every trap is red, even when something is cleared according to the logs).

EVENT oraEM4Alert .1.3.6.1.4.1.111.15.2.0.1 "Status Events" Normal
FORMAT The variables included in the oraEM4Alert trap. $*
SDESC
The variables included in the oraEM4Alert trap.
Variables:
  1: oraEM4AlertTargetName
     Syntax="OCTETSTR"
     Descr="The name of the target to which this alert applies."
... (lines removed for brevity)
  8: oraEM4AlertSeverity
     Syntax="OCTETSTR"
     Descr="The severity of the alert e.g. Critical."
... (lines removed for brevity)
EDESC


I was rereading the snmptt documentation, and it looks like I might be able to use the "match' rule in the file.  I can give that a try.


Thanks,
Nicole


From: FARRIOR, Andy [mailto:Andy.Farrior (at) victoriacollege.edu]
Sent: Wednesday, November 03, 2010 2:32 PM
To: xymon (at) xymon.com
Subject: [xymon] RE: help setting up SNMP traps to Xymon

The severity of the trap is defined in the corresponding snmptt.conf file.  That's what makes the determination if the trap is Normal, SEVERE, or whatever.

The only way to change the trap status on Xymon for a given device is for Xymon to receive a new trap (or the aforementioned fictional tool to acknowledge or set-to-normal the trap status) from the device that's different than the current status.

To do that, something has to happen to create an event on that device for a "Normal" SNMP trap to be sent.  That would replace whatever trap status Xymon shows for that device and change the status to "green".

Example:

2010-10-30 15:56:11

.1.3.6.1.4.1.318.0.9

[cid:image001.gif@01CB7C25.83D9E300]INFORMATIONAL

APC UPS: Utility power restored: Returned from battery backup power; utility power restored.

2010-10-30 15:56:05

.1.3.6.1.4.1.318.0.5

[cid:image002.gif@01CB7C25.83D9E300]WARNING

APC UPS: On battery: The UPS has switched to battery backup power.


In this instance the device sent an "WARNING" level event and it followed up with an "INFORMATIONAL" event that changed the trap status to "green".

If your device doesn't send a "back to normal" type of trap, I'm suggesting you could generate an event on that device that would send a "Normal" or "INFORMATIONAL" level trap.  You can look through the snmptt.conf file for your device and find these traps and try to figure out what you could do to generate that trap.  My example of saving new copy of a configuration on that device may not generate a trap at all.  Just have to find something that does.

But like I said, it's an ugly workaround ....


Sorry if I confused you.  Hope this helps,
Andy



From: Nicole Beck [mailto:nskyrca (at) syr.edu]
Sent: Wednesday, November 03, 2010 9:02 AM
To: xymon (at) xymon.com
Subject: [xymon] RE: help setting up SNMP traps to Xymon

Thanks Andy.

I thought that the severity of the trap came from what is defined in the /etc/snmp/snmptt.conf.oracle file?  The cerebro.victoriacollege.edu website has a section saying:

I've noticed that the APC and Dell MIB files have a SEVERITY definition in them. SNMPTT uses that to establish the severity for each trap (Normal|INFORMATIONAL|SEVERE|WARNING|...). However, I've noticed that Cisco and Canoga Perkins don't have those definitions; so, every trap event is considered Normal. You'll need to change the severity for the various traps as desired in the snmptt.conf file.

Does the snmptt.conf.oracle file have to be changed in order to deal with the different severity codes that it receives from the snmptrap?  It seems like the way my snmptt.conf.oracle file is configured right now, it would always  have a severity of "Normal", so Xymon would never change colors.

I'm just learning SNMP and don't completely understand it yet, so maybe I'm missing something.


I'll have to check with my DBA's again to find out if they can send their traps directly from the database server, rather than thru the Grid server.


Thanks,
Nicole



From: FARRIOR, Andy [mailto:Andy.Farrior (at) victoriacollege.edu]
Sent: Friday, October 15, 2010 12:48 PM
To: xymon (at) xymon.com
Subject: [xymon] RE: help setting up SNMP traps to Xymon

1 - clearing the trap.

Unfortunately, the only way the SNMP trap status changes on Xymon is when it receives a new SNMP trap from the device.  If the device would send an "all clear" type of trap when it was happy, that'd change the status back to green.  If there's a way to create an event on the device that generates a "Normal" event, you could use that as a mechanism to change the SNMP status on Xymon.   (Example, if saving the configuration generates a "Normal" SNMP trap....)  Ugly, but it'd work.

The other way to address it would be to come up with a way to manually acknowledge or clear a trap similar to the alert acknowledgment in Xymon.  I've never had time to pursue that.

2 - origination of the trap

The only way I can think of addressing this is if each database server can be configured to send SNMP traps on their own.  The SNMPTRAPD daemon uses the sending device's hostname and that's what is used to send to Xymon.
Hope this helps,
Andy



From: Nicole Beck [mailto:nskyrca (at) syr.edu]
Sent: Friday, October 15, 2010 9:42 AM
To: 'xymon (at) xymon.com'
Subject: [xymon] help setting up SNMP traps to Xymon

Hello,
I'm running Xymon 4.2.3 on RHEL 5.4 and I'm trying to setup Xymon to receive SNMP traps from our Oracle database servers.    We are also running Oracle Enterprise Manager on another server (we call it the "grid") that monitors the Oracle database servers too, but our DBA's want to use Xymon to alert them of Oracle problems instead of the grid server so they don't get alerts from two different places.   The DBA's manage OEM(therefore I'm not familiar with it), and have configured it to recieve SNMP traps and send them to Xymon.

I configured the Xymon server to receive traps using the documentation at http://cerebro.victoriacollege.edu/hobbit-trap.html.  We do receive traps, but there are two problems:


 1.  The trap status diamond does not change color when it gets a trap or when the trap clears.  My understanding is that this is set in the Oracle MIBS that are translated to the /etc/snmp/snmptt.conf.oracle file.  I'm not sure how to change the file to get it to give different alerts.  If I change the word "NORMAL" in the following line to "SEVERE" the trap status will go red, but will not return to green when the trap clears. Any suggestions on how to fix this?

EVENT oraEM4Alert .1.3.6.1.4.1.111.15.2.0.1 "Status Events" Normal


 1.  The traps status changes for the grid server instead of the database server with the problem.  Drilling down into the "trap" status, you will see an error for the database server though.   The snmptrapd.log and snmptt.log files (examples below) show the traps coming from the grid server.  Does anyone know how to get the traps to appear under the database server name rather than the grid server?

Example snmptrapd.log entry (host name changed):
2010-10-14 19:04:49 grid-server [xxx.xxx.xxx.xxx] (via UDP: [xxx.xxx.xxx.xxx]:51617) TRAP, SNMP v1, community public
        .1.3.6.1.4.1.111.15.2 Enterprise Specific Trap (1) Uptime: 0:06:31.00
        .1.3.6.1.4.1.111.15.1.1.1.2.1 = STRING: "DB00157.world" .1.3.6.1.4.1.111.15.1.1.1.3.1 = STRING: "Database Instance"     .1.3.6.1.4.1.111.15.1.1.1.4.1 = STRING: "database-server "   .1.3.6.1.4.1.111.15.1.1.1.5.1 = STRING: "Status".1.3.6.1.4.1.111.15.1.1.1.6.1 = ""      .1.3.6.1.4.1.111.15.1.1.1.7.1 = ""     .1.3.6.1.4.1.111.15.1.1.1.8.1 = STRING: "Oct 14, 2010 7:04:00 PM EDT"    .1.3.6.1.4.1.111.15.1.1.1.9.1 = STRING: "Critical"      .1.3.6.1.4.1.111.15.1.1.1.10.1 = STRING: "Failed to connect to database instance: ORA-01033: ORACLE initialization or shutdown in progress (DBD ERROR: OCISessionBegin)."       .1.3.6.1.4.1.111.15.1.1.1.11.1 = STRING: "Test SNMP database-server DB Down "       .1.3.6.1.4.1.111.15.1.1.1.12.1 = STRING: "SYSMAN"       .1.3.6.1.4.1.111.15.1.1.1.13.1 = STRING: "0"    .1.3.6.1.4.1.111.15.1.1.1.14.1 = ""     .1.3.6.1.4.1.111.15.1.1.1.15.1 = STRING: "923A730174F36B87E040E6801417642D"     .1.3.6.1.4.1.111.15.1.1.1.16.1 = STRING: "0"    .1.3.6.1.4.1.111.15.1.1.1.17.1 = ""     .1.3.6.1.4.1.111.15.1.1.1.18.1 = STRING: "0"    .1.3.6.1.4.1.111.15.1.1.1.19.1 = ""     .1.3.6.1.4.1.111.15.1.1.1.20.1 = STRING: "1"    .1.3.6.1.4.1.111.15.1.1.1.21.1 = STRING: "923A421065A8A060E040E68015173993"

Example snmptt.log entry:
Oct 14 19:04:52 xymon-server snmptt[0]: .1.3.6.1.4.1.111.15.2.0.1 SEVERE "Status Events" grid-server - The variables included in the oraEM4Alert trap. DB00157.world Database Instance database-server Status   Oct 14, 2010 7:04:00 PM EDT Critical Failed to connect to database instance: ORA-01033: ORACLE initialization or shutdown in progress (DBD ERROR: OCISessionBegin). Test SNMP database-server DB Down  SYSMAN 0  923A730174F36B87E040E6801417642D 0  0  1 923A421065A8A060E040E68015173993



Thanks in advance for any help.  I'm sorry for the lengthy question.

Nicole Beck