Prometheus新手求助：SNMP Exporter监控自定义扩展OID异常

Hey there! Let's work through getting your custom extend OID properly monitored with Prometheus. It sounds like you've got the basics set up, but there are a few common snags that might be tripping you up. Let's break this down step by step:

1. Double-Check You're Targeting the Right OID #

First off—you might be scraping the wrong metric from the NET-SNMP-EXTEND-MIB. The nsExtendResult OID you're checking is actually the exit code of your script (0 for success, non-zero for errors), not the output of your echo $VALUE command.

To get the actual output from your script, you need to target nsExtendOutputFull instead. Run this command to find its OID:

snmpwalk -v2c -c public localhost NET-SNMP-EXTEND-MIB::nsExtendOutputFull."value-return-test" -On

That's the OID you should be configuring in SNMP Exporter, not the result code.

2. Configure SNMP Exporter to Scrape the Correct Metric #

Update your snmp.yml configuration file to include a module for your custom extend OID. Replace the example OID below with the one you got from the nsExtendOutputFull snmpwalk:

modules:
  custom_value_return:
    walk:
      - .1.3.6.1.4.1.8072.1.3.2.4.1.2.16.118.97.108.117.101.45.114.101.116.117.114.110.45.116.101.115.114
    metrics:
      - name: value_return_test_output
        oid: .1.3.6.1.4.1.8072.1.3.2.4.1.2.16.118.97.108.117.101.45.114.101.116.117.114.110.45.116.101.115.114
        type: gauge  # Use "string" if your output is text, but Prometheus handles numeric metrics better
        help: "Output from custom value-return-test script"

Restart the SNMP Exporter service after making this change.

3. Verify SNMP Exporter Can Fetch the Metric #

Test if the SNMP Exporter is correctly pulling the metric by running this curl command (replace <snmp-exporter-ip> with your exporter's IP or use localhost if it's on the same machine):

curl http://<snmp-exporter-ip>:9116/snmp?target=localhost&module=custom_value_return

Look through the output for the value_return_test_output metric. If it shows up with your expected value, great—move on to checking Prometheus. If not, double-check your OID in snmp.yml and make sure the SNMP Exporter can reach the snmpd service.

4. Ensure Your Script Works for the SNMP User #

The snmpd daemon runs as the snmp user by default, so you need to make sure this user can execute your script and that the script outputs the correct value.

• Set proper permissions on your script:

  chmod +x /usr/local/bin/script.sh
  chown snmp:snmp /usr/local/bin/script.sh
  

• Test running the script as the snmp user:

  su - snmp -s /bin/bash -c "/usr/local/bin/script.sh"
  

If this doesn't output your expected $VALUE, fix the script's logic or permissions before proceeding.

5. Update Prometheus to Scrape the New Metric #

Add a scrape job to your prometheus.yml to pull metrics from the SNMP Exporter's custom module:

scrape_configs:
  - job_name: 'snmp_custom_extend'
    static_configs:
      - targets: ['<snmp-exporter-ip>:9116']
    params:
      module: ['custom_value_return']

Restart Prometheus, then head to the Prometheus UI and search for value_return_test_output—you should see your metric data there.

内容的提问来源于stack exchange，提问作者Skullone