NaN values from poller, values from manual operations

[rasekm]

Hello

I am collection statistical information using SNMP queries, and I have noticed that when using the aggregated graph that if a device reboots or is unreachable from Cacti. The aggregated graph stops updating. I am assuming this is because there is no data present as the polling is failing.

Is there anyway I can set a "failure" value in this particular case? Or will I have to create a script which does these sort of health checks?

[rasekm]

I created a PHP script in accordance to the Script Server layout. But when I am using it one of the parts of the graph responds with a "NaN". I have tried the same argument outside of Cacti and I am in fact getting a non-NaN value back.

I am even getting a NaN value when I hardcode the PHP script to return a value (i.e. 5) no matter what I am asking of the script.

If I run a "realtime" polling of the specific device I am getting back the hardcoded and correct values. But for some reason in the aggregated graph I am not getting any other values said device other than NaN.

Any ideas?

[rasekm]

My interpretation from the debug statement bellow is that the CMDPHP inserts values for dev2 when PHPSVR haven't done the polling of dev2. How can I solve this?

08/31/2018 11:07:35 AM - PHPSVR: Poller[0] PHP Script Server has Started - Parent is cmd
08/31/2018 11:07:35 AM - PHPSVR: Poller[0] DEBUG: PID[19057] CTR[0] INC: 'ss_ip_sla.php' FUNC: 'ss_ip_sla' PARMS: 'dev1 294 2:161:500:1:10:public:::::: get pkt_loss 1'
08/31/2018 11:07:35 AM - CMDPHP: Poller[0] DEVEL: SQL Exec: "insert into poller_output_rt (local_data_id, rrd_name, time, poller_id, output) values (32386, 'dev1_pkt_loss', '2018-08-31 11:07:35', '37809', '0')"
08/31/2018 11:07:35 AM - CMDPHP: Poller[0] DEVEL: SQL Exec: "insert into poller_output_rt (local_data_id, rrd_name, time, poller_id, output) values (32385, 'dev2_pkt_loss', '2018-08-31 11:07:35', '37809', 'U')"
08/31/2018 11:07:35 AM - PHPSVR: Poller[0] DEBUG: PID[19057] CTR[0] RESPONSE:''
08/31/2018 11:07:35 AM - PHPSVR: Poller[0] DEBUG: PID[19057] CTR[1] INC: 'ss_ip_sla.php' FUNC: 'ss_ip_sla' PARMS: 'dev2 2560 2:1612:30:public:::::: get pkt_loss 1'

[RustedKnight]

If you have changed the script and in particular the way the variables are output from the script into your graph template then try ripping out the graph and rebuilding it.

[rasekm]

If you have changed the script and in particular the way the variables are output from the script into your graph template then try ripping out the graph and rebuilding it.

I have done this numerous times, I can get the "NaN" values to be fixed very temporarily by issuing a realtime polling of the NaN value in the graph. However this tactic doesn't hold longer than 30seconds at a time.

This is my script server script

Code: Select all

function ss_ip_sla($hostname, $host_id, $snmp_auth, $cmd, $arg1 = "", $arg2 = "") {
    $snmp = explode(":", $snmp_auth);
	$snmp_version 	= $snmp[0];
	$snmp_port    	= $snmp[1];
	$snmp_timeout 	= $snmp[2];
	$ping_retries 	= $snmp[3];
	$max_oids		= $snmp[4];

	$snmp_auth_username   	= "";
	$snmp_auth_password   	= "";
	$snmp_auth_protocol  	= "";
	$snmp_priv_passphrase 	= "";
	$snmp_priv_protocol   	= "";
	$snmp_context         	= "";
	$snmp_community 		= "";

	if ($snmp_version == 3) {
		$snmp_auth_username   = $snmp[6];
		$snmp_auth_password   = $snmp[7];
		$snmp_auth_protocol   = $snmp[8];
		$snmp_priv_passphrase = $snmp[9];
		$snmp_priv_protocol   = $snmp[10];
		$snmp_context         = $snmp[11];
	}else{
		$snmp_community = $snmp[5];
	}

	$oids = array(
		"pkt_loss",
        "index"         => ".1.3.6.1.4.1.9.9.42.1.2.1.1.3",
        "tag"           => ".1.3.6.1.4.1.9.9.42.1.2.1.1.3",
        "vrf"           => ".1.3.6.1.4.1.9.9.42.1.2.2.1.26",
        "admintimeout"  => ".1.3.6.1.4.1.9.9.42.1.2.1.1.7"
		);

	if ($cmd == "index") {
		$return_arr = ss_ip_sla_index(cacti_snmp_walk($hostname, $snmp_community, $oids["index"], $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol, $snmp_priv_passphrase, $snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, $max_oids, SNMP_POLLER));

        foreach($return_arr as $index => $value){
            print_r($index."\n");
        }

	}elseif($cmd == "admin_tag"){
        $indexes = ss_ip_sla_index(cacti_snmp_walk($hostname, $snmp_community, $oids["tag"], $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol, $snmp_priv_passphrase, $snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, $max_oids, SNMP_POLLER));
        foreach($indexes as $oid => $value){
            print $oid . "!" . $value . "\n";
        }
    }elseif($cmd == "vrf"){
        $indexes = ss_ip_sla_index(cacti_snmp_walk($hostname, $snmp_community, $oids["vrf"], $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol, $snmp_priv_passphrase, $snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, $max_oids, SNMP_POLLER));
        foreach($indexes as $oid => $value){
            print $oid . "!" . $value . "\n";
        }
    }elseif($cmd == "admintimeout"){
        $indexes = ss_ip_sla_index(cacti_snmp_walk($hostname, $snmp_community, $oids["admintimeout"], $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol, $snmp_priv_passphrase, $snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, $max_oids, SNMP_POLLER));
        foreach($indexes as $oid => $value){
            print $oid . "!" . $value . "\n";
        }    
    }elseif ($cmd == "num_indexes") {
		$return_arr = ss_ip_sla_index(cacti_snmp_walk($hostname, $snmp_community, $oids["index"], $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol, $snmp_priv_passphrase, $snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, $max_oids, SNMP_POLLER));

		print sizeof($return_arr);

	}elseif ($cmd == "query") {
        $arg = $arg1;

		$indexes = ss_ip_sla_index(cacti_snmp_walk($hostname, $snmp_community, $oids[$arg], $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol, $snmp_priv_passphrase, $snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, $max_oids, SNMP_POLLER));
        
        if($arg == "index"){
            foreach($indexes as $index => $value){
                print_r($index."!".$index."\n");
            }
        }else{
            foreach($indexes as $oid => $value){
                print $oid . "!" . $value . "\n";
            }
        }
        

	}elseif ($cmd == "get") {
		$arg = $arg1;
        $index = $arg2;

        if($arg == "pkt_loss"){
            $pkt_loss_sd    = cacti_snmp_get($hostname, $snmp_community, ".1.3.6.1.4.1.9.9.42.1.5.2.1.26" . ".$index", $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol,$snmp_priv_passphrase,$snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, SNMP_POLLER);
            $pkt_loss_ds    = cacti_snmp_get($hostname, $snmp_community, ".1.3.6.1.4.1.9.9.42.1.5.2.1.27" . ".$index", $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol,$snmp_priv_passphrase,$snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, SNMP_POLLER);
            $pkt_loss_mia   = cacti_snmp_get($hostname, $snmp_community, ".1.3.6.1.4.1.9.9.42.1.5.2.1.29" . ".$index", $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol,$snmp_priv_passphrase,$snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, SNMP_POLLER);
            $pkt_oos        = cacti_snmp_get($hostname, $snmp_community, ".1.3.6.1.4.1.9.9.42.1.5.2.1.28" . ".$index", $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol,$snmp_priv_passphrase,$snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, SNMP_POLLER);
            $pkt_la         = cacti_snmp_get($hostname, $snmp_community, ".1.3.6.1.4.1.9.9.42.1.5.2.1.30" . ".$index", $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol,$snmp_priv_passphrase,$snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, SNMP_POLLER);
            $num_rtt        = cacti_snmp_get($hostname, $snmp_community, ".1.3.6.1.4.1.9.9.42.1.5.2.1.1" . ".$index", $snmp_version, $snmp_auth_username, $snmp_auth_password, $snmp_auth_protocol,$snmp_priv_passphrase,$snmp_priv_protocol, $snmp_context, $snmp_port, $snmp_timeout, $ping_retries, SNMP_POLLER);


            #Sanity Check
            if(
                ($pkt_loss_sd == "") ||
                ($pkt_loss_ds == "") ||
                ($pkt_loss_mia = "") ||
                ($pkt_oos == "") ||
                ($pkt_la == "") ||
                ($num_rtt == "")
            ){
                print("1\n");
            }else{
                $cisco_ipsla_pkt_loss = ($pkt_loss_sd+$pkt_loss_ds+$pkt_loss_mia)/($pkt_loss_sd+$pkt_loss_ds+$pkt_loss_mia+$pkt_oos+$pkt_la+$num_rtt);
                $ipsla_responder_down = (1-($num_rtt/10));
                $downtime = $cisco_ipsla_pkt_loss + $ipsla_responder_down;
                print($downtime."\n");
            }
        }
	}
}

function ss_ip_sla_index($arr){
    $return_arr = array();

    foreach($arr as $a){
        preg_match("/\d*$/",$a["oid"],$matches);
        $return_arr[$matches[0]] = $a["value"];
    }

    return $return_arr;
}

[rasekm]

1535709900: -nan
1535710200: -nan
1535710500: -nan
1535710800: -nan
1535711100: -nan
1535711400: -nan
1535711700: -nan
1535712000: 0.0000000000e+00
1535712300: -nan
1535712600: 0.0000000000e+00
1535712900: 0.0000000000e+00
1535713200: -nan

This is a snippet from my rra file. I am seeing in the debug lines that the value U is inserted. But I cannot reproduce this code, I am always getting back a number (when I run the same commands from prompt).

[netniV]

Check the poller output cache, see what is calling this script for that device and run those exact commands as the user the poller is being run as.