On@sharkymark's cluster, the logstream pod seems to restart periodically. It still sends logs, but doesn't seem to run for a long time.

We were able to find some logs from GCP. The format isn't great but there are definitely errors which causes the container to exit and restart: