problems with snmp data polls

[dguerre]

Hello,

I'm running cacti version 0.8.8a on debian wheezy. I had some graphs showing cpu usage of some cisco switches and a fortinet firewall. This graphs have stopped working recently (the graphs show -nan). Upon investigation I think that the polls for these graphs are not running. What I mean is that if I run the poller for one of the hosts manually (php -q cmd.php xx xx) y can see this in the logs:

Code: Select all

09/04/2013 02:15:54 PM - CMDPHP: Poller[0] Time: 0.0305 s, Theads: N/A, Hosts: 1
09/04/2013 02:15:54 PM - CMDPHP: Poller[0] Host[47] DS[507] SNMP: v2: 10.81.0.10, dsname: traffic_out, oid: .1.3.6.1.2.1.2.2.1.16.10101, output: 117169870
09/04/2013 02:15:54 PM - CMDPHP: Poller[0] Host[47] DS[507] SNMP: v2: 10.81.0.10, dsname: traffic_in, oid: .1.3.6.1.2.1.2.2.1.10.10101, output: 1081288823
09/04/2013 02:15:54 PM - CMDPHP: Poller[0] Host[47] DS[508] SNMP: v2: 10.81.0.10, dsname: traffic_in, oid: .1.3.6.1.2.1.2.2.1.10.10102, output: 2242872489
09/04/2013 02:15:54 PM - CMDPHP: Poller[0] Host[47] DS[508] SNMP: v2: 10.81.0.10, dsname: traffic_out, oid: .1.3.6.1.2.1.2.2.1.16.10102, output: 1086673830
09/04/2013 02:15:54 PM - CMDPHP: Poller[0] Host[47] RECACHE DQ[1] OID: .1.3.6.1.2.1.1.3.0, output: 125823117
09/04/2013 02:15:54 PM - CMDPHP: Poller[0] Host[47] RECACHE DQ[1] OID: .1.3.6.1.2.1.1.3.0
09/04/2013 02:15:54 PM - CMDPHP: Poller[0] Host[47] RECACHE: Processing 1 items in the auto reindex cache for '10.81.0.10'.
09/04/2013 02:15:54 PM - CMDPHP: Poller[0] Host[47] SNMP: Host responded to SNMP 

Those are the polls for interface traffic, but the one for cpu usage never shows up.

I tried rebuilding the poller cache, but that didn't solve the problem. I have also verified the snmp queries using snmpwalk and snmpget and those work correctly. I think there is a bug somewhere or I'm missing something.

Any help will be appreciated.

[gandalf]

If a rebuild poller cache does not help, the data sources seem to have been lost. They usually are the base for creating new entries for the poller_cache
R.

[dguerre]

If a rebuild poller cache does not help, the data sources seem to have been lost. They usually are the base for creating new entries for the poller_cache
R.

I have tried deleting and recreating the data sources and that doesn't work either.

[gandalf]

Please see "System Utilities -> View Poller Cache" again. Are the expected items missing?
R.

[dguerre]

Please see "System Utilities -> View Poller Cache" again. Are the expected items missing?
R.

Those items are in the poller cache, but are not being graphed:

looks like this

poller cache.PNG (3.41 KiB) Viewed 1669 times

[gandalf]

Please run

Code: Select all

SELECT *
FROM `poller_item`
WHERE host_id =47
AND `arg1` = ".1.3.6.1.4.1.9.9.109.1.1.1.1.8.1"

from either phpMyAdmin or mysql command line and post results.

I need this, cause the display you've posted does not contain all data required for analysis. Again, you will want to hide the credentials

R.

[dguerre]

Please run

Code: Select all

SELECT *
FROM `poller_item`
WHERE host_id =47
AND `arg1` = ".1.3.6.1.4.1.9.9.109.1.1.1.1.8.1"

from either phpMyAdmin or mysql command line and post results.

I need this, cause the display you've posted does not contain all data required for analysis. Again, you will want to hide the credentials

R.

this is the result:

Code: Select all

mysql> SELECT * FROM `poller_item` WHERE host_id =47 AND `arg1` = ".1.3.6.1.4.1.9.9.109.1.1.1.1.8.1";
+---------------+-----------+---------+--------+---------+------------+----------------+--------------+---------------+---------------+--------------------+----------------------+--------------------+--------------+-----------+--------------+----------+-------------------------------+---------+----------+---------------+----------------------------------+------+------+
| local_data_id | poller_id | host_id | action | present | hostname   | snmp_community | snmp_version | snmp_username | snmp_password | snmp_auth_protocol | snmp_priv_passphrase | snmp_priv_protocol | snmp_context | snmp_port | snmp_timeout | rrd_name | rrd_path                      | rrd_num | rrd_step | rrd_next_step | arg1                             | arg2 | arg3 |
+---------------+-----------+---------+--------+---------+------------+----------------+--------------+---------------+---------------+--------------------+----------------------+--------------------+--------------+-----------+--------------+----------+-------------------------------+---------+----------+---------------+----------------------------------+------+------+
|           557 |         0 |      47 |      0 |       1 | 10.81.0.10 | redactedcommunity       |            2 |               |               |                    |                      |                    |              |       161 |          500 | 5min_cpu | /var/lib/cacti/rra/47/557.rrd |       1 |      300 |           120 | .1.3.6.1.4.1.9.9.109.1.1.1.1.8.1 |      |      |
+---------------+-----------+---------+--------+---------+------------+----------------+--------------+---------------+---------------+--------------------+----------------------+--------------------+--------------+-----------+--------------+----------+-------------------------------+---------+----------+---------------+----------------------------------+------+------+
1 row in set (0.00 sec)

Thanks for your help

[gandalf]

Did you fiddle around with 60 sec polling and/or the data template? At least the rrd_next_step is worrying me. As a try, please update this value to "0" instead of "120".
R.

[dguerre]

Did you fiddle around with 60 sec polling and/or the data template? At least the rrd_next_step is worrying me. As a try, please update this value to "0" instead of "120".
R.

How do I do that, in the settings? or directly on the database?

[gandalf]

Directly in the database, please
R.

[dguerre]

Directly in the database, please
R.

I made the change, and it worked for a couple of polls but then the value on the database went back to 120. Maybe I need to change the value elsewhere?

[gandalf]

Directly in the database, please
R.

I made the change, and it worked for a couple of polls but then the value on the database went back to 120. Maybe I need to change the value elsewhere?

Yep. I recommend visiting the data template.
R.

[dguerre]

Directly in the database, please
R.

I made the change, and it worked for a couple of polls but then the value on the database went back to 120. Maybe I need to change the value elsewhere?

Yep. I recommend visiting the data template.
R.

What do I change in the data template?

[gandalf]

Step and heartbeat have to be taken care of. STEP must match the polling interval, heartbeat is usually chosen twice the STEP. See explanation in the rrdtool documentation to understand, why.
In case you need more help, please post data template screenshot.
And, most important: those changes WON'T propagate into existing rrd files. You will have to use rrdtool tune to change any existing rrd file
R.

[dguerre]

Step and heartbeat have to be taken care of. STEP must match the polling interval, heartbeat is usually chosen twice the STEP. See explanation in the rrdtool documentation to understand, why.
In case you need more help, please post data template screenshot.
And, most important: those changes WON'T propagate into existing rrd files. You will have to use rrdtool tune to change any existing rrd file
R.

Attached is the screenshot of the data template I'm using.

data template.PNG (52.13 KiB) Viewed 1612 times