What do the metrics windows_os_info, node_boot_time_seconds and node_network_up tell you, and in what form?

windows_os_info is an info metric: its value is always 1 and the facts (product, version, build) sit in its labels. node_boot_time_seconds is the last boot as a Unix timestamp. node_network_up is 1 or 0 per network interface, depending on whether its operational state is "up".

• windows_os_info{product="Microsoft Windows Server 2022 Standard", version="10.0.20348", build_number="20348", …} 1. The info-metric pattern stores text, which Prometheus values can't hold, in labels and sets the value to 1. Useful for joining OS details onto other series.
• node_boot_time_seconds is seconds since 1 January 1970 (Unix epoch). Convert it to a date, or compute the uptime with time() - node_boot_time_seconds.
• node_network_up{device="eth0"} is 1 when the interface's operstate is "up", else 0. One series per interface, including loopback and virtual ones, so an alert should filter the devices it cares about.

Finding metrics: the expression field in Prometheus's Graph view autocompletes, and the Metrics Explorer lists everything. All Windows Exporter metrics start with windows_, all Node Exporter metrics with node_ (e.g. windows_cs_logical_processors for the logical CPU count; newer exporter releases call it windows_cpu_logical_processor).

Go deeper:

• windows_exporter — os collector — the labels of windows_os_info with an example line.
• Unix time — Wikipedia — the seconds-since-1970 format node_boot_time_seconds uses.