Traces are not delivered after a few days of runtime

[BlitzInternet]

using docker image opentracing/nginx-opentracing:0.26.0-alpine
with a config:

{
	"service_name": "nginx",
	"diabled": false,
	"reporter": {
		"logSpans": true,
		"endpoint": "http://obs:14268/api/traces"
	},
	"sampler": {
		"type": "const",
		"param": 1
	}
}

the problem: after several days of flawless operation, the spans are no longer sent to the collector endpoint.
the trace-ids are still generated and passed on correctly, which causes the a warning of "" in tempo when trying to view the trace.
not a single span is received then.
once that period starts, the solution we have to use is to re-create the docker container.

[miry]

@BlitzInternet Can you create a script with docker compose, how can we reproduce it.