• Bugfix: possible incorrect values of metrics in Prometheus output that used variables other than $p8s_value for their values; in practice the issue could occur with angie_http_upstreams_peers_state and angie_stream_upstreams_peers_state from the standard prometheus_all.conf template.
• Bugfix: some connection attempts to upstream servers might not have been properly accounted for in the statistics API if they failed immediately; the bug had appeared in 1.3.0.