Skip to content

Releases

v3.3.0 — Major performance overhaul: 8x faster exports, real multi-core parallelism

Choose a tag to compare

@asinay asinay released this 29 Sep 03:26

SysPerfSight

This release is headlined by a major performance overhaul: large report exports that used to hang indefinitely (or take multiple minutes) now reliably finish in seconds, with real multi-core parallelism and a live progress indicator. The rest is a batch of reliability and UX fixes accumulated since v3.2.0.

Performance

  • Exports on large reports no longer hang indefinitely — parsing, analysis, and output generation now run through a thread pool with timeouts, so a pathological file fails cleanly with a 504 instead of blocking the server forever
  • Per-section analyzers now run in real separate worker processes instead of threads, giving genuine multi-core parallelism (analyzers are pure CPU work with nothing to await, so threads alone mostly serialize on the GIL) — cross-section synthesis also now runs concurrently with the analyzers instead of waiting for them
  • Fixed a severe iostat slowdown on systems with many disk devices (multipath LUNs, lots of local disks): a report with 75 devices was generating 528 individual Plotly chart-trace calls plus repeated full-dataframe scans — iostat: 111s → 6.6s, total export: 116s → 14.7s (~8x), output size: ~121MB → ~85MB on the same real-world file. Charts now cap to the top 15 devices by %util (readable, and the summary table/insights still cover every device)
  • mgstat, iostat, sar-d, and %SS gained the same row/interval/snapshot downsampling cap already used by vmstat/perfmon, bounding chart size on long captures
  • Export progress is now visible: a live "Generating report… Ns elapsed (N sections, X MB)" status shows under the export buttons instead of just a static "Generating…" button label
  • Per-stage timing is logged server-side (docker logs) for future diagnosis without needing the report file itself — see PERFORMANCE.md

Fixes

  • Uploads no longer silently fail on files re-saved by a browser ("Save Page As" instead of the raw source) — this reserializes the section-heading markup into a structurally different but visually identical form; it's now detected and normalized automatically
  • A 0-section parse now returns an actionable 400 error (with a diagnostic hint) instead of a false "success" with an empty section list
  • Error messages are now readable even when a reverse proxy rejects a request before it reaches the app (e.g. a 413 from an upload size limit) instead of an opaque Unexpected token '<' from trying to parse an HTML error page as JSON
  • Fixed the Docker image still referencing the pre-rename pbuttons_parser.py
  • Renamed the published GHCR image from sysperf-sight to sysperfsight to match the repo/container name

UI

  • Added a FAQ page and made all internal fetch/link URLs relative, so the app keeps working when reverse-proxied under a subpath
  • The generated report's header now matches the app's own header style (wordmark, colors, dark-mode toggle)
  • General UI redesign as an ops tool, with leftover pButtons-era branding cleaned up

v3.2.0 — Caché/Windows compatibility, chart thresholds, and IRIS-aware disk insights

Choose a tag to compare

@asinay asinay released this 09 Aug 14:23

SysPerfSight

This release fixes several real bugs found while testing against an older Windows Caché report, and adds three enhancements inspired by reviewing similar community tools (yape/yaspe).

Fixes

  • %SS now parses correctly on Caché reports — previously produced no output at all, silently, because the snapshot regex required an IRIS-specific header prefix
  • Caché's differently-named sections (license, cpf file, ccontrol all, cstat -D/-c1) are now correctly flagged sensitive and routed to their analyzers, instead of being silently skipped
  • irisstat -D's "Per-Second Rates" table now renders — a header-parsing bug meant it had never rendered for anyone, on any platform
  • perfmon now tracks each physical disk separately instead of silently discarding all but one drive's data — on a multi-disk Windows system this could mean charting and alerting on the wrong disk entirely
  • restart.bat fixed after the project folder rename (and the underlying venv console-script shim gotcha documented, so it won't be a mystery next time)

New

  • Dashed threshold reference lines on sar-d/iostat/sar-u/vmstat/perfmon charts (disk %util, CPU busy, latency), matching each chart's own insight thresholds
  • Disk-saturation insights now cross-reference the CPF file's declared database/journal/WIJ paths against the real disk layout (Linux mount, Windows drive letters), so warnings scope to IRIS-relevant disks instead of firing on unrelated devices
  • Insight "peak" values use the 99th percentile instead of raw max, reducing false alarms from single-sample spikes

v3.1.0 — Renamed to SysPerfSight

Choose a tag to compare

@asinay asinay released this 26 Jun 15:15

SysPerfSight

This release completes the rename from pButtons Parser to SysPerfSight — reflecting the tool's expanded scope and the report's updated name (InterSystems SystemPerformance, formerly pButtons).

Changes

  • App title and UI branding updated to SysPerfSight throughout
  • Parser module renamed: pbuttons_parser.py → sysperfsight_parser.py
  • Output filenames now use sysperf_ prefix
  • Docker image: ghcr.io/asinay/sysperf-sight
  • All internal references, docstrings, and placeholder text updated

No functional changes — all features from v3.0.0 are unchanged.

v3.0.0 — Export modes, insight markers, 5 new analyzers

Choose a tag to compare

@asinay asinay released this 26 Jun 14:40

What's new

Export modes

Three buttons replace the single "Download filtered file":

  • Full report — charts, insights, raw data, cross-section synthesis, sensitive banners
  • Charts + Raw — charts + raw data (collapsed); no insights or synthesis
  • Charts only — charts only; excluded sections hidden entirely

Output filename is now auto-filled as pbuttons_YYYYMMDD_HHMM.html on upload, with _charts_raw / _charts suffixes appended for the lighter modes.

Insight markers

All analyzers wrap their insights block in <!--INS-->…<!--/INS--> markers so the reduced export modes can strip insights with a single regex — stat cards and charts are always retained.

New analyzers

Section What it produces
df -m Stacked disk usage bar chart; colour-coded Use% table; capacity insights
ps Process snapshot: RSS by user, top-15 by RSS, IRIS job-type breakdown; D-state insights
mount Local and network filesystem tables; soft-mount and NFS sync insights
irisstat -R Routine buffer pool: in-use routines, top packages, type breakdown; LRU eviction insights
Synthesis Cross-section LLM summary panel (Full report mode only)

UI

  • Filename field on its own row above the buttons; all three buttons always on one row
  • "Full report" replaces "Download filtered file"

v2.1.0 — vmstat, sar-u, irisstat-D analyzers and sortable tables

Choose a tag to compare

@asinay asinay released this 30 May 16:56

What's new

New analyzers

  • vmstat — Run queue + blocked processes, swap I/O, CPU breakdown (user/sys/id/wa/st), block I/O charts; stat cards; insights flagging sustained run queue, high swap activity, and CPU saturation
  • sar -u — Stacked area CPU chart (user/sys/iowait/steal with idle as dotted line); stat cards from the Average: line; insights for CPU saturation, steal, and iowait
  • irisstat -D — Lock contention summary table + per-second rates table (both sortable); insights flagging Bseize > 0 on Global/LockHTAB/LockLHB/TransCB; block collision flags; handles Linux (4-col) and Windows (6-col) column variants

vmstat time-range filtering

vmstat rows are now filterable by time window alongside mgstat, iostat, sar -u, and sar -d — scoping all charts and aggregates to the period of interest.

Sortable tables in all analyzers

A shared sortTable() utility is now injected into every exported HTML file, making all analyzer summary tables (irisstat-D, sar-d device summary, %SS process tables, etc.) sortable by any column with ascending/descending toggle.

Minor fixes

  • Sections with injected analysis are no longer pre-collapsed even if listed in COLLAPSED_BY_DEFAULT
  • Analyzer exceptions now print full tracebacks to the server console for easier debugging

All processing is 100% local — no data leaves your machine.

v2.0.0 — Analyzer system, time filters, and expanded section coverage

Choose a tag to compare

@asinay asinay released this 30 May 16:46

What's new

Inline analysis with Plotly charts

Each section now shows charts and insight flags injected above its raw data. Analyzers ship for:

  • mgstat — Glorefs, PhyRds/Wrs, Journal writes, GblSz, BytSnt/Rcd, and more (4×2 chart grid + stat cards)
  • iostat — %util, CPU iowait, IOPS, throughput, latency
  • sar -d — %util, tps, throughput, r/w latency, queue depth; per-device summary table
  • %SS — Process type breakdown, TCP trend, top-CPU/Glob tables, per-namespace and per-routine tables
  • cpu — CPU topology summary
  • tasklist — Top processes by memory (Windows)
  • Windows info — OS/hardware summary cards

Time-range filtering

Filter mgstat, iostat, sar -u, and sar -d to a specific time window before exporting. Instead of analyzing the full 24-hour dataset, you can scope all charts and aggregates to just the period of interest — making peaks, averages, and trends far more meaningful.

Expanded section support

Full section coverage for Linux pButtons (vmstat, sar -u, sar -d, iostat, cpu, free, lsblk, etc.) and Windows pButtons (perfmon, tasklist, Windows info).

Sidebar navigation groups

Sections are grouped in the output sidebar (IRIS Core, Storage, Network, OS, Windows).


All processing is 100% local — no data leaves your machine.

pButtons parser v1

Choose a tag to compare

@asinay asinay released this 29 May 23:00