Data Watch 1.4.0: Memory observability

Data Watch now observes memory alongside network traffic. Open the new Memory tab or http://127.0.0.1:8765/#memory to inspect RAM headroom, swap, paging, Linux memory pressure, kernel OOM-kill counters, and programs ranked by resident memory.

The existing Rust service samples every six seconds independently of the network connection. It keeps 30 days of memory history: system summaries by minute and top-program peaks in five-minute windows. Select a timeline bar to inspect recorded peaks, or Return to live. Program rows open details; the list starts with ten groups and expands for scrolling.

No runtime dependencies or privileged helper changes are required. Existing network history remains intact. RSS can double-count shared memory; pressure and low-headroom indicators are heuristics, and the monitor does not stop programs or change cache/swap settings. Earlier freezes cannot be reconstructed.

A short 20-second measurement on the development PC observed 5.5 MiB RSS for the service versus 5.2 MiB for the previous version. CPU was 2.9% versus 2.3% of one core, including waited child processes. This excludes the browser and packet helper and is not a long-term benchmark.

Validation: 24 Rust tests, clippy with warnings denied, JavaScript syntax checks, network and memory UI controller checks, standalone binary HTTP/assets/notifications/memory smoke checks, installer checks, and installed memory persistence plus network-capture reconnection checks passed. Browser automation was unavailable; visual layout and native browser interactions remain unverified.

Download the x86-64 Ubuntu archive, verify SHA256SUMS, extract, and run ./data-watch install. Installation opens the local dashboard in a graphical session. The README lists Linux requirements and measurement limits.