gui: add snapwr to boot_progress, fix live_network_metrics docs

[jherrera-jump]

No description provided.

[Copilot]

Pull request overview

This PR extends Firedancer’s GUI boot progress reporting to include snapwr (snapshot write) stage metrics, and updates API/docs to reflect the expanded boot progress payload and live network metrics layout.

Changes:

• Add snapwr_accounts_written gauge and track it in the snapwr tile.
• Include snapwr progress (in-bytes, out-bytes, accounts) in boot_progress summary output and websocket docs.
• Update live_network_metrics documentation to include the rserve protocol slot.

Reviewed changes

Copilot reviewed 7 out of 9 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
src/discof/restore/fd_snapwr_tile.c	Track accounts_written and preserve full-snapshot baseline across incremental/retries.
src/disco/metrics/metrics.xml	Define the new Snapwr AccountsWritten gauge.
src/disco/metrics/generated/fd_metrics_snapwr.h	Generated metadata for new Snapwr gauge.
src/disco/metrics/generated/fd_metrics_snapwr.c	Generated metric table includes new Snapwr gauge.
src/disco/gui/fd_gui.h	Extend boot progress snapshot struct with snapwr fields.
src/disco/gui/fd_gui.c	Sample snapwr metrics and compute per-snapshot (full vs incremental) deltas.
src/disco/gui/fd_gui_printf.c	Emit new boot progress JSON fields for snapwr.
book/api/websocket.md	Document new boot progress fields and fix live network metrics examples/tiles list.
book/api/metrics-generated.md	Generated metrics docs include new Snapwr gauge.

[greptile-jt]

Greptile Summary

This PR adds snapshot write (snapwr) stage metrics to the GUI boot progress display and updates the websocket API documentation.

• Adds a new AccountsWritten gauge metric to the snapwr tile, with a full_accounts_written backup field to correctly split cumulative counts between full and incremental snapshot phases
• Surfaces three new snapwr fields in the boot progress JSON output: snapwr_in_bytes_decompressed, snapwr_out_bytes_decompressed, and snapwr_accounts
• Applies per-snapshot baseline subtraction to insert_accounts (fixing a pre-existing issue where the incremental snapshot showed cumulative counts instead of per-snapshot counts)
• Documents the rserve (repair serve) network protocol in the websocket API docs and updates live_network_metrics examples to reflect the current 6-protocol set

Confidence Score: 4/5

This PR is safe to merge; it adds observability metrics and documentation with correct baseline subtraction logic.

The new snapwr metrics follow established patterns from neighboring snapshot tiles. The baseline subtraction logic is correct and consistent with the existing cumulative metric handling. The only minor concern is that the websocket docs imply EMA arrays have the same element count as raw ingress/egress arrays, but this is a pre-existing issue.

book/api/websocket.md has a minor documentation accuracy issue with EMA array element counts.

Important Files Changed

Filename	Overview
src/disco/gui/fd_gui.c	Adds snapwr tile metrics lookup and per-snapshot baseline subtraction for snapwr and insert accounts. Logic correctly mirrors the existing snapin/snapdc patterns with saturating subtraction.
src/disco/gui/fd_gui.h	Adds three new fields (snapwr_in_bytes_decompressed, snapwr_out_bytes_decompressed, snapwr_accounts_current) to the boot_progress loading_snapshot struct. Clean, consistent with existing members.
src/disco/gui/fd_gui_printf.c	Adds JSON output fields for the new snapwr boot progress metrics in both the active and null branches of the HANDLE_SNAPSHOT_STATE macro. Consistent with existing output patterns.
src/discof/restore/fd_snapwr_tile.c	Adds accounts_written metric with full_accounts_written backup mechanism for full/incremental snapshot transitions. Correctly increments on account header processing and manages resets on INIT/NEXT/FAIL control messages.
src/disco/metrics/metrics.xml	Adds AccountsWritten gauge metric to the snapwr tile definition.
src/disco/metrics/generated/fd_metrics_snapwr.h	Generated file updated to include ACCOUNTS_WRITTEN offset and metadata. Total metric count correctly updated to 4.
src/disco/metrics/generated/fd_metrics_snapwr.c	Generated file updated to declare the new ACCOUNTS_WRITTEN gauge metric.
book/api/websocket.md	Documents new snapwr boot progress fields and rserve network protocol. The EMA array examples suggest 6 elements but code only outputs 5 (rserve missing from EMA).
book/api/metrics-generated.md	Auto-generated metrics documentation updated to include the new snapwr AccountsWritten metric.

Sequence Diagram

sequenceDiagram
    participant snapct as snapct tile
    participant snapdc as snapdc tile
    participant snapin as snapin tile
    participant snapwr as snapwr tile
    participant gui as GUI (fd_gui)
    participant ws as WebSocket Client

    Note over snapwr: INIT_FULL: reset accounts_written=0
    snapwr->>snapwr: Process full snapshot accounts
    snapwr->>snapwr: Increment accounts_written per account header
    Note over snapwr: NEXT: backup full_accounts_written = accounts_written

    Note over snapwr: INIT_INCR: accounts_written = full_accounts_written
    snapwr->>snapwr: Process incremental snapshot accounts
    snapwr->>snapwr: Increment accounts_written per account header

    loop Every 100ms sampling
        gui->>snapct: Read FULL_SIZE_BYTES, FULL_BYTES_READ, etc.
        gui->>snapdc: Read FULL_DECOMPRESSED_BYTES_WRITTEN, etc.
        gui->>snapin: Read ACCOUNT_LOADED (cumulative)
        gui->>snapwr: Read FULL_BYTES_READ, ACCOUNTS_WRITTEN, BYTES_WRITTEN
        gui->>gui: Compute per-snapshot values via baseline subtraction
        gui->>ws: Broadcast boot_progress JSON (incl. snapwr fields)
    end

Loading

_{Reviews (1): Last reviewed commit: "gui: add snapwr to summary.boot_progress..." | Re-trigger Greptile}

[greptile-jt]

EMA arrays shorter than documented

The ingress_ema and egress_ema arrays will have 5 elements (turbine, gossip, tpu, repair, metric) because FD_GUI_NET_PROTO_CNT is 5 and rserve is excluded from the EMA rate calculation in fd_gui_network_rate_max_update (fd_gui.c:643-655). Meanwhile, the ingress and egress arrays output 6 elements (including rserve, at fd_gui_printf.c:726/734).

Showing 5 example values + ... for all four arrays suggests they all have the same length, which would be confusing for consumers who discover the mismatch at runtime.