|
|
| Author |
Message |
pashen
Helper
Joined: 12 May 2004
Posts: 22
|
 Posted: Sat Jun 05, 2004 1:50 pm Post subject: I'm not able to add snmp test or edit snmp test. |
 |
|
When I had netmrg 0.14 I've deleted some unnecessary snmp tests. Now with 0.15 and 2004.06.05 from cvs when I choose "Add" in "SNMP Tests" I see proper form. I enter information about new snmp test (Name, SNMP OID ...) and click "Save changes" . Then I see:
DB_ERROR: Sorry, a database error occured. We cannot continue. Please contact the administrator and let them know what you were doing when the problem occured
Regards.
Pashen |
|
| Back to top |
|
 |
silfreed
Site Admin
Joined: 19 Jan 2003
Posts: 394
Location: Shippensburg, PA, USA
|
| Posted: Sat Jun 05, 2004 2:42 pm Post subject: |
|
|
I'm not sure what's going on with your 0.15 release, but if you're running current CVS you have to maintain the database changes manually.
When you're running from CVS, make sure to execute the items in the DBChangeLog. We recently made some changes to SNMP tests to support Windows boxen better, but had to make some changes to the database structure. The updater usually handles these changes, but only from release to release (or pre-release or release candidate), not to CVS.
-Doug |
|
| Back to top |
|
|
pashen
Helper
Joined: 12 May 2004
Posts: 22
|
| Posted: Sat Jun 05, 2004 9:02 pm Post subject: |
|
|
| silfreed wrote: |
if you're running current CVS you have to maintain the database changes manually.
When you're running from CVS, make sure to execute the items in the DBChangeLog. We recently made some changes to SNMP tests to support Windows boxen better, but had to make some changes to the database structure. The updater usually handles these changes, but only from release to release (or pre-release or release candidate), not to CVS.
-Doug |
Thanks. I update database manually and everything looks good. (Now I use cvs version).
I''ve such case:
From cisco via snmp I get value which isn't pure integer:
[L: 0064, C: 0016] [Dev: 0009] [Sub: 0043] [Mon: 0117] Value: 278000 bits per second
[L: 0128, C: 0128] [Dev: 0009] [Sub: 0043] [Mon: 0117] RRD: 'update /var/lib/netmrg/rrd/mon_117.rrd N:278000 bits per second'
[L: 0064, C: 0016] [Dev: 0009] [Sub: 0043] [Mon: 0117] Value is non-integer.
Will you change netmrg-gatherer source to send to RRD not whole Value but only this part which is integer ?
Of course I can use workaround: perl script which will get this value from cisco , regexp and add this script to "Test scripts'.
But I hope you will add such "feature".
Regards.
Pashen |
|
| Back to top |
|
|
balleman
Site Admin
Joined: 20 Jan 2003
Posts: 282
|
| Posted: Sun Jun 06, 2004 4:38 pm Post subject: |
|
|
| We can look into that. It is really rare to have units included in the response to an SNMP get. What OID are you querying? For most things that are "bits per second" we usually just graph the counter (and specify that as the monitor type). |
|
| Back to top |
|
|
pashen
Helper
Joined: 12 May 2004
Posts: 22
|
| Posted: Sun Jun 06, 2004 11:29 pm Post subject: |
|
|
| balleman wrote: |
| We can look into that. It is really rare to have units included in the response to an SNMP get. What OID are you querying? For most things that are "bits per second" we usually just graph the counter (and specify that as the monitor type). |
.1.3.6.1.4.1.9.9.166.1.15.1.1.11.9999.9999 where 9999.9999 are policy-map and class-map indexes.
Pashen |
|
| Back to top |
|
|
silfreed
Site Admin
Joined: 19 Jan 2003
Posts: 394
Location: Shippensburg, PA, USA
|
| Posted: Sun Jun 06, 2004 11:59 pm Post subject: |
|
|
Okay, so you're getting this all from CISCO-CLASS-BASED-QOS-MIB.my, and the one you're trying to graph is cbQosCMPostPolicyBitRate; I haven't done too much QoS on Ciscos (or read this mib very much), but it looks like cbQosCMPostPolicyByte64 (.1.3.6.1.4.1.9.9.166.1.15.1.1.10.x.x) should be the same value, just with a better unit (and 64 bit counters). I could be reading the mib wrong, but maybe graph it along-side the other one and see?
-Doug |
|
| Back to top |
|
|
pashen
Helper
Joined: 12 May 2004
Posts: 22
|
| Posted: Mon Jun 07, 2004 7:13 am Post subject: |
|
|
You read MIB right. I know there are counters. Reading PolicyBitRate has one advantage. It provides current value. With counters you have one problem.
When you read them after longer time than five minutes you will see peak on graph. Such situation occur from time to time. That's why I'd like to read MIB returning "current" values.
Pashen |
|
| Back to top |
|
|
|
