Skip to content

Silence noise

For a message you know is noise: one that repeats, or a stack trace you've seen a hundred times. Lowering the logger's level would hide everything else it says too. A rule removes just that message, or keeps it and shortens its stack trace. See Rules for how rules work.

WildFly only in 1.0

Rules need java.util.logging or JBoss LogManager. On a plain JVM with Logback they aren't available yet.

Above: an exception flooding a WildFly console. top finds the logger, and a drop rule silences that exception while the rest of the application keeps logging.

The output on this page is from WildFly 26.1.3.

Drop a message

Name the logger and something in the message:

logctl add rule drop io.undertow.request --message-contains "Matched default" --below WARN --reason "health checks"
r2   io.undertow.request → drop   (FOR, reverts 05:27:09 local — in 3h 59m)

--below WARN limits the rule to events below WARN. Without it, the rule covers everything below ERROR. A drop rule needs something to match besides the level; to quiet a whole logger, use set logger with a higher level instead.

The rule lets the first matching event through, then one every 5 minutes, so the full message is never far away in the log. Every dropped event is counted.

Shorten a stack trace

Above: a payment endpoint is down, and every order logs the same ConnectException with its full stack trace. A trim rule keeps every warning but cuts that exception to one line, so the rest of the console is readable again.

To keep an event but not its trace:

logctl add rule trim '*.Worker' --throwable java.io.IOException --frames 3 --yes
r3   org.logaperture.sample.work.Worker → trim   (FOR, reverts 05:27:09 local — in 3h 59m)

Each matching IOException is still logged, with its top three frames and a marker saying how many were cut. Without --frames, the exception is one line. A trim rule needs no matcher: logctl add rule trim com.acme.Worker --below WARN shortens every trace from that logger below WARN.

The '*.Worker' pattern adds one rule to every logger now matching it. On a terminal, logctl lists them to pick from; --yes takes all of them.

Matching on the exception

Option Matches
--message-contains <text> the log message contains the text
--message-contains-ignore-case <text> the same, ignoring case
--throwable <class> the exception is exactly this class
--throwable-message-contains <text> the exception's message contains the text
--any-cause let the two above match any exception in the cause chain

Give several and all must match.

See and change rules

logctl list rules --verbose
ID  LOGGER                              EXPRESSION                                                          ACTION  TIER  EXPIRES               HITS
r2  io.undertow.request                 --message-contains "Matched default" --below WARN --sample-full 5m  drop    FOR   07:19:12 (in 3h 59m)  0
r3  org.logaperture.sample.work.Worker  --throwable java.io.IOException --below ERROR --frames 3            trim    FOR   07:19:13 (in 3h 59m)  0

HITS counts the events each rule has acted on. EXPRESSION is the rule's options, in the form alter rule takes. To change a rule without removing it, give only what changes:

logctl alter rule r2 --below ERROR
logctl alter rule r2 sticky

To remove one:

logctl reset rule r2
rule r2 → removed.

reset rules removes all of them, except sticky ones unless you add --include-sticky.

Let logctl ask

On a terminal, logctl add rule alone walks through every part: which logger (starting from part of its name, such as Deployer), drop or trim, what to match, and how long. It shows the full command before adding the rule. logctl alter rule r2 alone shows the rule's parts to pick what to change.