Non-error logs are printed to stderr

[ku1ik]

As in title. This makes alerting configuration based on log severity problematic because everything seems to be an error. Most logging aggregation tools and platforms treat stderr output as level=error.

For example, "New tick of statistics from 10.9.1.100: 2497 ads blocked / 26870 total DNS queries" is not an error but it logged to stderr.

Steps for Reproduction

Run the exporter, redirect stderr to a file, observe all logs being saved there.

Expected behavior:

Informational logs are printed to stdout, and only errors are printed to stderr.

Actual behavior:

All logs, including informational, are printed to stderr.

Platforms:

Ubuntu 20.04, x86 64-bit

Versions:

0.1.1

[Galorhallen]

This should be handled with #110, which also introduce log levels

[ku1ik]

Nice!

[wzzrd]

@proggaras Did it though? Because I still get all logs through stderr even today. I was just browsing through my logs, an all logs generated by pihole exporter - even though they say level=info in the body - are still logged to stderr, not stdout.

Maybe I'm missing something though?

[wzzrd]

~/tmp/2024-08-21
➜ ./pihole_exporter-darwin-amd64 > foo
INFO[0000] ------------------------------------
INFO[0000] -  PI-Hole exporter configuration  -
INFO[0000] ------------------------------------
INFO[0000] Go version: go1.20.1
INFO[0000] PIHoleProtocol : [http]
INFO[0000] PIHoleHostname : [127.0.0.1]
INFO[0000] PIHolePort : [80]
INFO[0000] Port : 9617
INFO[0000] Timeout : 5s
INFO[0000] ------------------------------------
INFO[0000] New Prometheus metric registered: domains_blocked
INFO[0000] New Prometheus metric registered: dns_queries_today
INFO[0000] New Prometheus metric registered: ads_blocked_today
INFO[0000] New Prometheus metric registered: ads_percentag_today
INFO[0000] New Prometheus metric registered: unique_domains
INFO[0000] New Prometheus metric registered: queries_forwarded
INFO[0000] New Prometheus metric registered: queries_cached
INFO[0000] New Prometheus metric registered: clients_ever_seen
INFO[0000] New Prometheus metric registered: unique_clients
INFO[0000] New Prometheus metric registered: dns_queries_all_types
INFO[0000] New Prometheus metric registered: reply
INFO[0000] New Prometheus metric registered: top_queries
INFO[0000] New Prometheus metric registered: top_ads
INFO[0000] New Prometheus metric registered: top_sources
INFO[0000] New Prometheus metric registered: forward_destinations
INFO[0000] New Prometheus metric registered: querytypes
INFO[0000] New Prometheus metric registered: status
INFO[0000] Creating client with config <Config@C000282058 PIHoleProtocol=http, PIHoleHostname=127.0.0.1, PIHolePort=80>
INFO[0000] Starting HTTP server
^C2024/08/21 10:41:05 pihole-exporter HTTP server stopped

~/tmp/2024-08-21
➜ cat foo

~/tmp/2024-08-21
➜ ./pihole_exporter-darwin-amd64 > foo 2>&1

~/tmp/2024-08-21
➜ head foo
time="2024-08-21T10:41:10+02:00" level=info msg=------------------------------------
time="2024-08-21T10:41:10+02:00" level=info msg="-  PI-Hole exporter configuration  -"
time="2024-08-21T10:41:10+02:00" level=info msg=------------------------------------
time="2024-08-21T10:41:10+02:00" level=info msg="Go version: go1.20.1"
time="2024-08-21T10:41:10+02:00" level=info msg="PIHoleProtocol : [http]"
time="2024-08-21T10:41:10+02:00" level=info msg="PIHoleHostname : [127.0.0.1]"
time="2024-08-21T10:41:10+02:00" level=info msg="PIHolePort : [80]"
time="2024-08-21T10:41:10+02:00" level=info msg="Port : 9617"
time="2024-08-21T10:41:10+02:00" level=info msg="Timeout : 5s"
time="2024-08-21T10:41:10+02:00" level=info msg=------------------------------------

As you see above, when I redirect output with > foo, the file "foo" remains empty, because stdout is not being used. Only when I redirect stderr to stdout the "foo" file is filled with messages.