We have about 5,000 users at the moment. We get roughly 25 of these errors a day and are yet to receive any support requests about issues relating to this. We can’t really run the troubleshooting information because of this.
The nginx log has no errors being posted and we’re just using the default nginx configuration. The fact it’s working in the vast majority of cases (>99.9%) I’m happy for us to put it down to network instability / similar.