This is an old revision of the document!

FastDPI monitoring and logs

System logs are presented as text files that are located in the /var/log/dpi directory for DPI and PCRF modules. Types of messages in the log:

[CRITICAL] - critical error, system operation is impossible without troubleshooting
[WARNING] - warning, the system does not stop, but it is worth eliminating this malfunction
[TRACE] - messages when the diagnostic trace mode is enabled
[INFO] - notification of system actions
[ERROR] - error when connecting services and policies, incorrect configuration

The FastDPI process by default logs all system actions to the following debug and statistics log files:

/var/log/dpi/fastdpi_slave.log - a log of traffic processing processes¹⁾
/var/log/dpi/fastdpi_stat.log - traffic processing statistics log
/var/log/dpi/fastdpi_alert.log - common fastDPI functions log

Blocking counters that are saved in the statistics log

Log file rotation

File rotation provides a daily backup of the daily log. By default, this process is performed during the hours with the lowest system load. The log storage depth is defined in the configuration of /etc/logrotate.d/fastdpi by the parameter maxage, the value is specified in days.

Monitoring SNMPD

Description

SNMP is a protocol that allows retrieving information about the DPI server for centralized monitoring of critical parameters.
How it works: an SNMP request is sent to the server, it retrieves the necessary information and sends it to the monitoring server. The monitoring server receives and processes this information, allowing to understand the server's status—tracking at which levels various parameters and data are located.

The bngsnmp package contains scripts needed to form the VENDOR branch of OIDs for SNMP polling.
It is based on the snmp_passpersist library. Vendor branch OID: .1.3.6.1.4.1.43823

Contents of the vendor branch:

CPU statistics, including core utilization by BNG processes;
number of illegitimate drops;
drop statistics on the dispatcher;
availability list of RADIUS servers (when using a proxy or load balancer, this proxy will be shown in the statistics);
number of DHCP subscribers — when the function is active;
list of VRFs and active subscribers in each — when the function is active;
DPDK interface statistics (number of packets, errors, signal level, etc.);
NAT statistics (list of profiles, list of pools, and number of translations for each public address) — when the function is active;
list of hardware modules installed on the motherboard — if supported by the BMC controller;
metrics from hardware sensors (voltage, power consumption, FAN RPMs) — if supported by the BMC controller;
utilization of DHCP server pools (supported only by KEA-DHCP server) — when the function is active.

This script also overrides standard branches by adding statistics from DPDK interfaces:

.1.3.6.1.2.1.2.2.1 — 32-bit counters
.1.3.6.1.2.1.31.1.1.1 — 64-bit counters

If necessary, you can disable the override by commenting out or deleting the following lines from /etc/snmp/snmpd.conf:

pass_persist    .1.3.6.1.2.1.2.2.1      /usr/local/bin/bng_snmp/vas_pp.py
pass_persist    .1.3.6.1.2.1.31.1.1.1   /usr/local/bin/bng_snmp/vas_pp_hc.py

MIB File

Installation

Installation:
```
yum install bngsnmp
```

Configuration:

/bin/cp -f /etc/snmp/snmpd.conf.example /etc/snmp/snmpd.conf

Restart the net-snmp service:
```
systemctl restart snmpd
```
Disable selinux (in case of error when starting SNMP):
```
vi /etc/selinux/config
...
SELINUX=disabled
...
```

Checking SNMPD and vendor branch functionality

The snmpwalk utility allows you to check whether data is being received via SNMP:

Install snmpwalk:
```
dnf install net-snmp-utils
```

Command to check SNMP functionality:

snmpwalk -v 2c -c nokiamon localhost -On .1.3.6.1.4.1.43823

How to read the MIB file

If you need to see the MIB file in a graphical representation, you can open it through the MIB Browser. It will describe the entire vendor branch and its OIDs.
The example below shows:

Vendor branch file
Its OID

You can also create your own walk in the MIB Browser by specifying the server address

Known limitations

On-stick device support is available starting from version 13.2-beta4.2.

Monitoring via SNMP agent (Zabbix-agent)

We offer you the following set of parameters that can be taken from the SSG DPI:

Errors in fastDPI process log /var/log/dpi/fastdpi_alert.log
Errors in the /var/log/messages system log
Losses (Drop) on dna interfaces
Traffic volume on interfaces
Availability of control interfaces
Number of HTTP and HTTPS requests processed
Number of blocked resources by HTTP, HTTPS, IP
Number of PPPoE sessions

You can use Zabbix Agent for monitoring.
Current and final supported version of agent and server is 6.0, Zabbix agent 1 should be used. For newer versions of Zabbix, monitoring will be done via SNMP.

Agent Setup

Install Zabbix agent 1 on the DPI server according to the instructions on the Zabbix website.
In the first step, select the following values:
- Zabbix Packages
- Zabbix version: 6.0+
- OS distribution: CentOS
- OS version: 8 STREAM
- Zabbix component: AGENT
Edit the configuration file /etc/zabbix/zabbix_agentd.conf: change the parameters Server= and ServerActive= to your server address, hostname= to the server hostname.

Change the context of the /var/log/dpi/fastdpi_stat.log file:

chcon unconfined_u:object_r:zabbix_log_t:s0 /var/log/dpi/fastdpi_stat.log

Open tcp/udp ports 10050 and 10051 in firewall
Upload the ssg_userparams.conf file to the /etc/zabbix/zabbix_agent.d/ directory
Edit the ssg_userparams.conf file by replacing the interface number in UserParameter
02-00.0 should be replaced with the interface names of your server!
The name must match the DPI config. If you have more than 2 interfaces in use, you must add a line similar to the existing parameters.
```
UserParameter=dpi.02-00.0.drops,tac /var/log/dpi/fastdpi_stat.log | sed /'IF 02-00.0'/q | tac | sed -e 1,/'Actual   Stats'/d | sed '6!D' | awk '{print $1}' | sed 's/^.//'
```
Restart the agent: systemctl restart Zabbix-agent

Server setup

Install and configure Zabbix servers according to the instructions on the official website.
Add template zbx_export_templates.xml
1) Go to Configuration section
2) Templates section
3) Click “Import”
4) Import template file
5) Save changes
Add a DPI server as a host
1) Go to Monitoring section
2) Hosts section
3) Click “Create host”
4) Set the required parameters, hostname, group and the previously added template
5) Save changes
Edit the template: change the names of the interfaces and keys so that they match the UserParameter.

Monitoring traffic distribution by class

SSG allows traffic distribution by class to be monitored.

1. Enable traffic prioritization. For the example, we will use the following prioritization rules:

dns cs0
http cs0
https cs0
Bittorrent cs7
ICMP cs0
TCP Unknown cs7
GOOGLEVIDEO cs1
default cs2

2. In the /etc/dpi/fastdpi.conf configuration, set the parameter:

dbg_log_mask=0x4

3. Enable common channel polysync (the example shown is polysync with full channel width restriction):

htb_inbound_root=rate 1300mbit 
htb_inbound_class0=rate 8bit ceil 1300mbit
htb_inbound_class1=rate 8bit ceil 1300mbit
htb_inbound_class2=rate 8bit ceil 1300mbit
htb_inbound_class3=rate 8bit ceil 1300mbit
htb_inbound_class4=rate 8bit ceil 1300mbit
htb_inbound_class5=rate 8bit ceil 1300mbit
htb_inbound_class6=rate 8bit  ceil 1300mbit
htb_inbound_class7=rate 8bit  ceil 1300mbit
htb_root=rate 1300mbit 
htb_class0=rate 8bit ceil 1300mbit
htb_class1=rate 8bit ceil 1300mbit
htb_class2=rate 8bit ceil 1300mbit
htb_class3=rate 8bit ceil 1300mbit
htb_class4=rate 8bit ceil 1300mbit
htb_class5=rate 8bit ceil 1300mbit
htb_class6=rate 8bit  ceil 1300mbit
htb_class7=rate 8bit  ceil 1300mbit

4. Update the configuration:

service fastdpi reload

If polyscing for a shared channel is applied for the first time, you must restart the service:

service fastdpi restart

5. Use the following custom settings for the zabbix agent installed on the SSG: ssg_userparams.conf

6. Import the template to the Zabbix server as described in the section "Monitoring via SNMP agent": zbx_export_templates.xml

If necessary, change the interface names in the template and in the custom parameter file

View flow and protocol statistics

By flow

IPv4/IPv6
protocol type: 0 - IPv4, 1 - IPv6
total allocated records
a queue with a short lifespan:
1. occupied records
2. reusable
3. difference 3.1 - 3.2 (number of active flows)
also for the long line
also total

Example:

fdpi_ctrl stat --flow
IPv4 0 6784000 834 814 20 0 0 0 834 814 20

By protocols

internal index of protocol statistics
protocol name
protocol port number
direction subs -→ inet
number of packages
volume in bytes ip total
dropped packages
dropped byte
direction inet -→ subs number of packages etc.