Thanks Brian. I already use the regex_extracts for the SNMP floating point conversion via - https://www.robustperception.io/numbers-from-displaystrings-with-the-snmp_exporter - so yes this is what I was looking for (and overlooking).
also Thanks @sup - I had not thought about adding a recording rule, which has a bunch advantages in my use case. I will lookin further into. Cheers! On Saturday, October 23, 2021 at 12:42:12 AM UTC-7 [email protected] wrote: > The best way to do this is to use a recording rule. > > Something like this: > > - record: global:pduInputPowerConsumption:sum > expr: | > sum(pduInputPowerConsumption{job="apc_pdus"} * 10 or on () vector(0)) > + > sum(pduInputPowerConsumption{job!="apc_pdus"} or on () vector(0)) > > This is also nicer for your dashboard, since it only needs to pull one > metric from the TSDB. This will be much faster for long-term views. like " > avg_over_time(global:pduInputPowerConsumption:sum[$__interval])" > > I also recommend not trying to get clever with SNMP metric names. Just > keep the raw names generated with the snmp_exporter generator. This way you > can avoid the problems of mismatched metric issues. Different OIDs have > different meanings from different vendors. Keep them separated and handle > things at query time. Either in the dashboard or in your rules. > > On Sat, Oct 23, 2021 at 1:44 AM Ben Cohee <[email protected]> wrote: > >> This has been bothering me for a while, and hopefully someone has a >> solution I am simply overlooking. >> >> I use snmp_exporter to pull power metrics from a bunch of different PDU >> vendors (APC, Raritan, Geist, ServerTech, etc). I have one Prometheus >> scrape job & module per vendor to store all the PDU total power data in a >> metric named pduInputPowerConsumption >> >> All of these vendors report the power in Watts, except for APC - which >> reports in " hundredths of kilowatts (decawatts)" >> >> So throughout all my Grafana dashboards where I have to aggregate the >> total power consumption from all the different PDU vendors I currently just >> multiply the APC metrics by 10 to convert to watts like the other PDUs. >> >> sum(pduInputPowerConsumption{job="apc_pdus"}*10 OR on() vector(0)) >> + >> sum(pduInputPowerConsumption{job*!=*"apc_pdus"} OR on() vector(0)) >> >> >> Is there a way that I can just scale *10 the apc_pdu collected metric on >> the prometheus backend via the prometheus.yml scrape config along with >> my relabel_configs or metric_relabel_configs? >> Is there something else I am blatantly overlooking? >> >> >> >> >> # Scrape config for snmp_exporter polling APC PDU power metric >> pduInputPowerConsumption >> - job_name: 'apc_pdus' >> scrape_interval: 3m # Set the SNMP scrape inverval for every 3 mins >> scrape_timeout: 45s # Set the scrape timeout to 45 seconds (3 >> retries = 2.25 mins) >> metrics_path: /snmp >> params: >> module: [apc_pdus] >> file_sd_configs: >> - files: >> - /etc/prometheus/snmp-targets/apc_pdus_*.json >> # Attempt to re-read files every five minutes. >> refresh_interval: 5m >> relabel_configs: >> - source_labels: [__meta_filepath] >> regex: '.*_(\w{4})\.json' >> replacement: $1 >> target_label: site >> - source_labels: [__address__] >> target_label: ip_address >> - source_labels: [__address__] >> target_label: __param_target >> - target_label: __address__ >> replacement: 10.0.XX.XX:9116 # SNMP exporter. >> >> -- >> > 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 [email protected]. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/prometheus-users/7fbdaaa6-8aec-43d5-962b-8f4372075811n%40googlegroups.com >> >> <https://groups.google.com/d/msgid/prometheus-users/7fbdaaa6-8aec-43d5-962b-8f4372075811n%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> > -- 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 [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/fe3225d0-3e9f-4547-afcd-30f4dc62accen%40googlegroups.com.
