feat(ui): live disk/network throughput graphs + visible upload recovery - #290
Merged
Merged
Conversation
Two gaps the 2026-08-14 incident's follow-up bundle exposed: a resumed 88 GB
upload pushed 140 Mbps while every throughput surface read zero (the Activity
graph is an activity-log SQL window, and the resume writes no rows until it
completes), and the whole reconcile phase sat in an unlabeled indeterminate
"Starting backup..." sweep.
- driven-core: app-global cumulative IoCounters (plaintext disk-read bytes;
destination-acked wire bytes, credited exactly once per path - per ack for
resumable sessions incl. the resume, on completion for single-request
uploads; bundles by their session acks alone). New
OrchestratorState::Recovering carries the resume's path + byte totals;
Executor::reconcile takes a RecoverProgressSink, emitted at resume start
(before the prefix re-read produces any acks) and per ack, throttled ~1/s
in the orchestrator on the injected clock. The state-transition INFO log
now gates on the variant NAME so counter-carrying states no longer spam
(12 lines/60ms observed in the follow-up bundle).
- src-tauri: a 1s sampler task (updater-lifecycle pattern, quit-drained)
diffs the counters into a 5-minute ring, emits `sync:io_throughput`
(idle-suppressed after one trailing zero), and serves the ring via the new
`io_throughput_series` command. Recovering maps to the syncing bucket in
the tray.
- ui: the Activity header's throughput tile goes live (probe-fed 1s buckets)
and splits into Network + Disk read variants of one component; the global
progress bar renders Recovering as a DETERMINATE byte bar ("Recovering
interrupted upload - 8.2 GB of 88.6 GB"). Visual baselines regenerated.
v1 scope notes (documented in-code): scanner deep-verify hashing and restore
do not credit disk_read yet; bundle assembly reads are approximated by their
wire push; the adaptive ThroughputProbe deliberately still ignores resume
bytes (its pool-sizing semantics under an idle pool are unverified); the
post-resume adopt re-hash is a second full disk pass (tracked follow-up).
Tests: IoCounters accounting; resume emits monotonic recover ticks ending at
completion; resume credits disk (2x: re-read + adopt re-hash) and exactly the
unacked tail on the wire; sampler tick delta/ring/idle-suppression math;
iostat store seed/live/cap/rates; recovering phase precedence + determinate
percent; progress-bar recovering labels; disk tile variant; tile order. Full
workspace + 788 UI tests + 102 visual green.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A7q3CvJzL4zZmDA9CbXyQQ
…ks pass) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A7q3CvJzL4zZmDA9CbXyQQ
Contributor
Coverage
Gate: passed - no coverage regression (epsilon 0.1 pp). |
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A7q3CvJzL4zZmDA9CbXyQQ
pmaxhogan
added a commit
that referenced
this pull request
Aug 15, 2026
🤖 I have created a release *beep* *boop* --- ## [2.11.0](v2.10.1...v2.11.0) (2026-08-14) ### Features * **ui:** live disk/network throughput graphs + visible upload recovery ([#290](#290)) ([4fe156c](4fe156c)) ### Bug Fixes * **core:** exempt Zone.Identifier from the ads_skipped warning ([#288](#288)) ([f37ff68](f37ff68)) * let a slow add-source wizard finish instead of failing with a phantom disk error ([#291](#291)) ([1fbba98](1fbba98)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please).
pmaxhogan
added a commit
that referenced
this pull request
Aug 15, 2026
The v2.11.0 tag's release pipeline failed at the visual-regression gate (setup-wizard baselines predating #291) before publishing any artifacts, so no 2.11.0 build ever shipped. The baselines were regenerated in #292; this empty commit gives release-please a releasable change so 2.11.1 carries the full 2.11.0 payload (#288, #290, #291) through a green pipeline. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A7q3CvJzL4zZmDA9CbXyQQ
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two gaps the 2026-08-14 incident's follow-up diagnostics bundle exposed on v2.10.1: a resumed 88 GB upload pushed 140 Mbps while every throughput surface read zero (the Activity graph is an activity-log SQL window and the resume writes no rows until it completes), and the entire reconcile phase sat in an unlabeled indeterminate "Starting backup..." sweep.
What this adds
Live split throughput graphs. One app-global cumulative
IoCounters(driven-core) credited by every executor: plaintext disk-read bytes (upload reader stage, resume re-read, reconcile re-hash paths, buffered small-file band) and destination-acked network wire bytes (per ack for resumable sessions - including the resume, which previously credited nothing - and on completion for single-request uploads; bundles by their session acks alone, never double-counted). A 1s sampler task in the app shell (updater-lifecycle pattern, quit-drained, idle-suppressed after one trailing zero) diffs the counters into a 5-minute ring, served by the newio_throughput_seriescommand and pushed live viasync:io_throughput. The Activity header's throughput tile goes live and splits into Network throughput and Disk read variants of one component - both move in real time through every backup phase, including crash recovery.Visible upload recovery. New
OrchestratorState::Recovering { source_id, path, bytes_done, bytes_total }:Executor::reconcilegains aRecoverProgressSink(mirroringexecute'son_progress), emitted at resume start - before the prefix re-read produces any acks, so the UI flips out of the sweep immediately - and on every ack, throttled to ~1/s in the orchestrator on the injected clock (first and final ticks always pass; unit-tested). The global progress bar renders it as a determinate byte bar: "Recovering interrupted upload - 8.2 GB of 88.6 GB". Tray maps it to the syncing bucket.Log-noise fix ridealong: the state-transition INFO log now gates on the variant name, so counter-carrying states (Scanning/Recovering) no longer log every tick (12 lines in 60 ms observed in the follow-up bundle).
Known limitations (deliberate, documented in-code)
adopt_reconciled's P1-2 re-hash is a second full disk pass - for an 88 GB file the bar holds at 100% for a few minutes while it runs. The new disk graph shows exactly that activity, so the behavior is now legible; collapsing the second pass for the just-resumed case is a tracked follow-up.disk_readyet; bundle assembly reads are approximated by their wire push.ThroughputProbedeliberately still ignores resume bytes (its pool-sizing semantics against an idle upload pool are unverified).Progressevents) - cosmetic, macOS-only.Tests
Rust: IoCounters accounting; resume emits monotonic recover ticks ending at completion; resume credits disk (2x: re-read + adopt re-hash) and exactly the unacked tail on the wire; orchestrator throttle (5 stationary-clock ticks -> exactly first + final Recovering states); sampler tick delta/ring/idle-suppression. UI: iostat store seed/live/cap/rate math + garbled-wire coercion; recovering phase precedence + determinate percent; progress-bar recovering labels + aria; disk tile variant + default-unchanged; tile order. Full workspace suites, 788 UI unit tests, and all 102 visual tests (baselines regenerated in the same Docker harness and committed) are green locally.
README updated in this PR (Activity graphs, visible recovery, diagnostics-bundle contents).
Note for release sequencing: this is a
feat, so the next release is 2.11.0 (it also carries #288's Zone.Identifier fix).🤖 Generated with Claude Code
https://claude.ai/code/session_01A7q3CvJzL4zZmDA9CbXyQQ