Prometheus Scraping Abnormally

[Umut Cokbilir]

Hi All,

Do you know the root cause of the problem at attachment? Some devices display the interface data abnormally.

Device type: huawei router

prometheus version: prometheus-2.25.1.linux-amd64

scrape_interval: 2m scrape_timeout: 2m

snmp_exporter version: snmp_exporter-0.20.0.linux-amd64

snmp timeout:2m

Thanks

[Ben Kochie]

What is the query and query parameters for that graph?

It could either be due to a lost scrape, or a caching bug in the device's SNMP stack. I've seen that kind of thing when SNMP result caching is used on the device.

--
You received this message because you are subscribed to the Google Groups "Prometheus Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/fcfd370b-289a-4d2e-b52b-f8c075edcb82n%40googlegroups.com.

[Umut Cokbilir]

I haven't seen this problem in a previous version of snmp_exporter and prometheus.

Queries:

((rate(ifHCInOctets{ifAlias=~".*_NNI.*",ifDescr=~"[[int]]"}[10m]) * 8) and on (instance) sysName{sysName =~ ".*$name.*"})

((rate(ifHCOutOctets{ifAlias=~".*_NNI.*",ifDescr=~"[[int]]"}[10m]) * 8) and on (instance) sysName{sysName =~ ".*$name.*"})

6 Nisan 2021 Salı tarihinde saat 10:11:51 UTC+3 itibarıyla sup...@gmail.com şunları yazdı:

[Ben Kochie]

The version of Prometheus or snmp_exporter don't have anything to do with this. It's a data or scrape problem with the device.

If you do a console query for ifHCInOctets{ifAlias="X",ifDescr="Y",instance="Z"}[2h] at 22:00 for the series that you're seeing, it will return the raw data in the TSDB. That should help you debug what happened.

To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/5e8f3755-e0a0-4fb4-87eb-c1296f1b711bn%40googlegroups.com.