Buffered statsd lines are not published on close

[WillBrindle]

This may well be misunderstanding on my part but is there any way to ensure all counters are updated on close?
Here is a minimal example of the problem:

public static void main(String args[]) throws InterruptedException {
        StatsdConfig config = new StatsdConfig() {
            @Override
            public String get(String k) {
                return null;
            }
        };
        MeterRegistry registry = new StatsdMeterRegistry(config, Clock.SYSTEM);

        String METRIC = "test.metric";

        registry.counter(METRIC);
        registry.close();
}

If you spin up a simple udp server you'll see nothing is sent. This extends to if I add a sleep for a value more than step() , before or after closing. If I add a loop before closing where I increment the counter then sleep I start to get buffered messages across, but if not enough messages have been sent to clear the buffer by the last metrics are lost.

Setting buffered in my config to false resolves the problem but the natural behave feels like flushing should occur on close.

[mickaeltr]

Isn't it the same issue as #1882 ?

[gordon-rennie]

Isn't it the same issue as #1882 ?

I can still repro this issue on 1.11.0 and latest.integration builds, so sadly no; this issue is unfixed and outstanding. I noticed it because "job complete" metrics were not being emitted by a task kicked off by a k8s CronJob, causing alerting to go off.

[jonatan-ivanov]

#1882 fixed this for StepMeterRegistry implementations while StatsdMeterRegistry extends MeterRegistry.

[github-actions]

If you would like us to look at this issue, please provide the requested information. If the information is not provided within the next 7 days this issue will be closed.

[github-actions]

Closing due to lack of requested feedback. If you would like us to look at this issue, please provide the requested information and we will re-open.

[shakuzen]

Thanks to the pull request from @BJMg we now have tests that reproduce this behavior and changes to fix it, which are now merged into all supported branches and is published in snapshots. It will be included in future releases for 1.10.16, 1.11.13, 1.12.9, 1.13.3, and 1.14.0. Folks who were affected by this, please try out snapshots to let us know if the issue is fixed for you now.