Enormous 'One or more TimeSeries could not be written' stackdriver entries

[mieseprem]

Our services runs as GCP Cloud Run. We recently upgraded to Spring Boot 3.4 and Spring Cloud GCP 6.0 and since then an enormous amount of the following log entries are written:

failed to send metrics to Stackdriver
com.google.api.gax.rpc.InvalidArgumentException: io.grpc.StatusRuntimeException: INVALID_ARGUMENT: One or more TimeSeries could not be written: timeSeries[21]: Field timeSeries[21].points[0].distributionValue had an invalid value: Distribution value with |count| of 0 has a non-zero |mean| value of 276.26018395238094.; timeSeries[11]: Field timeSeries[11].points[0].distributionValue had an invalid value: Distribution value with |count| of 0 has a non-zero |mean| value of 246.; timeSeries[6]: Field timeSeries[6].points[0].distributionValue had an invalid value: Distribution value with |count| of 0 has a non-zero |mean| value of 3.0420689545454547.; timeSeries[15]: Field timeSeries[15].points[0].distributionValue had an invalid value: Distribution value with |count| of 0 has a non-zero |mean| value of 135.43589743589743

My knowledge here is too little as I can say if this is a duplicate of #4763. Maybe not, as we are not affected by issues arised after #4353 (in fact, it solved issues for us).

Environment

We run Java Spring Boot services using Java 21 in Linux based container with the following library versions:

• Spring Boot 3.4.3
• Spring Cloud GCP: 6.0.1
• Micrometer: 1.14.4 (registry Stackdriver)

To Reproduce
In our case it's just enable metrics export:

management.stackdriver.metrics.export.enabled = true

# we also set the following properties:
management.stackdriver.metrics.export.batch-size = 25

management.metrics.distribution.percentiles.http.server.requests = 0.99
management.metrics.tags.application = ${spring.application.name}
management.metrics.tags.instance = ${CLOUD_RUN_EXECUTION:local}
management.metrics.tags.random = ${random.uuid}

Expected behavior
All metrics are written to Stackdriver. With the current behaviour (of what is logged), I'm unsure if all metrics are written.

[mieseprem]

Oh, I've seen that you released 1.14.5 that addresses this issue already.

I will check if this is a duplicate of #5927.

[mieseprem]

Okay, I can confirm, that this issue is gone as of 1.14.5.

[jonatan-ivanov]

Duplicate of #5927