OpenSearch / Elasticsearch

• Detect cluster product and version from GET / (ES vs OS badge in header)
• OpenSearch: ISM explain instead of ILM; skip field usage stats and health report where unsupported
• Snapshot policies and some snapshot APIs gated for OS

Templates

• Three sections: Index templates, Legacy index templates, Component templates
• Detail modal supports component type; Rest recipes for template APIs

Slow indexing

• Fix false-positive merge stall alerts (no longer uses frozen total_time / current_size)
• 60s / 3-sample deep diagnosis with real block signals (merge stopped, write rejected, throttle, queue trend)
• CPU hot_threads from the correct suspect node

Rest

• Run multiple selected requests sequentially in the console

Nodes

• Network columns relabeled Incoming (RX) / Outgoing (TX)

Other

• Query icon on index names (Indexing & Search)
• Shard detail modal; cluster selector toolbar nowrap
• Install compare flow kept but disabled until 2027

Rest

• New Rest tab: recipes, history, confirm before write, curl copy, and Open in Rest from Query

Traffic

• Map and Active HTTP poll only _nodes/stats/http (no _cat/shards placement or _cat/nodeattrs)
• Group by index keeps full dated names (daily-history-2026-08-19) and patterns (daily-history-*)
• /consents/_doc/… groups as consents; cluster /_bulk stays in unknown
• Opening the unknown group shows only unknown URIs for those IPs (not mixed consents/bulk history)

Nodes

• Role pills, elected master, and Hot/Warm from _cat/nodes / HTTP roles
• Attributes column removed (_cat/nodeattrs no longer called)

Index / cluster

• Nested dialogs (Write task, Shard copy) stack above Index details; Escape/X close one layer
• Recovery translog stage shows Applying recent writes with ops progress

Index detail — Slow indexing

• Instant verdict from live write tasks and the write thread pool (no wait for the 15s sample)
• Refresh samples shard copies over 15s; copies table appears under a sampling banner
• Healthy empty copy: “No write stall” / write path looks healthy, without fake Watch banners
• In-flight write rows open a detail modal; node names resolve to Nodes tab correctly
• Longest bulk KPI uses a rounded duration (e.g. 2ms)

Index detail — Slow search

• Idle index shows All Clear! instead of zero KPI tiles

Query

• Time histogram Compare cycles off → 1d → 7d → 30d against the previous period

Index detail — Slow indexing diagnosis

• Surfaces long-running write/bulk tasks and shard placement as soon as the panel opens (no wait for the 15s sample)
• Node labels on shard samples and write tasks; clearer stall / stuck-node highlighting
• Hot threads when a stall is detected or a bulk has been running >60s
• Sample again keeps the previous report visible while resampling

Cluster API

• _cat/health support for cluster health context

UX

• Footer clock (local time)

Cluster — Active recovery

• Richer progress (%, left, average speed, ETA) and human-readable stages
• Row click opens detail modal (bytes/files/translog, hosts, snapshot context)
• Global Progress sort by % (sort full list, then paginate); default descending
• Silent 30s poll while the section is expanded

Snapshots

• Repositories and Snapshot policies sections (SLM + OpenSearch SM)
• Verify / execute / start-stop actions where applicable

Nodes

• Extended metric sections: disk, network, thread pools, circuit breakers, indexing failed
• Faster extended stats sampling (~10s)

UX

• Click-to-copy IPs: hover copy icon + check feedback (no link underline)
• Traffic tab moved to the far right
• Document flyout opens from the right

Cluster — Unassigned shards

• Fix allocation premium action with preview modal (retry_failed)
• Allocate replica to a preferred or other data node
• Actions column marked premium; quieter post-fix messaging
• Accordion layout no longer squeezes when multiple sections expand

Long-running tasks

• Cancel confirm uses an in-app modal instead of window.confirm
• Actions column labeled and marked premium

Diagnosis

• Node panel: stuck write path (hot threads, write tasks, sample)
• Index detail: Slow indexing tab (dual sample + stall detection)

Traffic & auth

• Hot/Warm/Cold/Frozen (+ multi-tier) role grouping on the map
• Clearer API key / Basic auth handling when browsers show native popups

Index detail

• Collapsible JSON tree for mappings, settings, and ILM explain (expand/collapse, search, copy)
• Mapping summary shows Primary fields and Multi-fields instead of text/keyword-only counts

Traffic

• Search and pagination in client activity and raw request history modals

Traffic tab

• Live Traffic map of clients → cluster roles (search / index / update / delete)
• Active HTTP traffic table with independent Group by (network, agent, index)
• Request history from poll deltas, including wildcard index patterns (e.g. daily-history-*)

Premium

• Signed license keys (trial / monthly / yearly) with paywall gates for Traffic and other premium surfaces
• Settings UI to activate a license; key generation script for maintainers

Also in this release

• Template detail modal and assorted Nodes / Indices / Query polish

Active searches — risk signals

• SGW-aligned detection for expensive search shapes (scripts, oversized aggs, huge terms lists, deep pagination, scroll)
• Reframed the old Pattern column as Risk to avoid implying root-cause

Query details — measured bottleneck

• Query Profile viewer for active search tasks
• Shows phase-level bottleneck (Query vs Aggregation vs Collect) so you can confirm where time is actually spent

Verification & tests

• Query profile verification script + timing audit
• Added unit tests for search classification and bottleneck inference

Query — index pattern picker

• Favorites section: built-in *, logs-*, metrics-* plus user-pinned targets (star)
• One-time top-3 suggestions from cluster stats (hidden/system indices excluded); re-computed only when all pins are cleared
• Indices / Aliases / Data streams collapsed by default; search expands matching sections
• Picker loads aliases and data streams on open

Query — time chart

• Fix empty histogram for naive date fields (yyyy-MM-dd HH:mm:ss) by omitting time_zone on non-ISO formats

Tables — sort

• Shared 3-state column sort (desc → asc → clear) with stable column widths
• Nodes tab: all columns sort by real values; Indices, Snapshots, Active searches updated

Cluster tab

• Long-running tasks section with grouped view, progress, cancel, and 30s expanded poll

Verification scripts

• Table sort, nodes table sort, query histogram alignment, query index favorites, cluster tasks

Query — Discover-style workspace

• Fields panel with Selected / Popular / Available / Empty / Meta groups
• Filter bar with match phrase and exclude (+ / -) actions from field top values
• Field top values popover with bar chart and quick filter actions
• Document flyout for row drill-down; cell action popover for filter/filter-out
• Meta columns, time field leftmost, natural index name sort

Query — time chart & probe

• Chart open probe: 15m → 24h → 30d → 1y → All; stops on first data
• Context-aware empty messages (range-specific text; All omits “extend time range”)
• Manual date field selection re-triggers search on indexes without @timestamp
• Index picker dropdown via portal (no clip from overflow)
• Chart loading states: full “Searching…” during probe; table kept stable via display latch

Query — table & columns

• Discover table layout and column defaults from field usage
• Auto-columns toggle persisted; improved sort/column reset on chart expand

Cluster & tabs

• Active searches refresh and task detail polish
• Cluster connection key dedupe across tabs

Verification scripts

• Chart probe, Discover filters/field groups/top values, index name sort, chart failover

Query — time chart

• Discover-style histogram: full window extended_bounds, auto interval, browser timezone
• Chart expand starts at 15m, falls back to All when empty
• Fixed bar layout for timezone-less date fields (e.g. yyyy-MM-dd HH:mm:ss)
• Horizontal axis labels; Kibana-style footer with range + interval
• Unified loading while field metadata and search run

Query — search performance

• track_total_hits no longer auto-enabled for single-index simple search (opt-in via Advanced JSON only)

Cluster — active searches

• Active searches section: filter, refresh, task detail modal, cancel with confirm

Index — diagnosis

• Diagnosis tab with batch run and conclusion UI
• Entry points from Indices table and Indexing & Search links

Nodes — investigate CPU

• Investigate CPU panel: thread pool + hot threads (throttled)

Alerts

• Deep links for high-CPU and slow-search alerts
• Thread pool metadata enrichment

UX & hardening

• Nested modal Escape stack (useNestedEscapeClose)
• Query index picker and chart expand cancel-storm fixes
• API gateway: search tasks, hot threads, index search stats endpoints

Query — time chart (Kibana-style)

• Time histogram with Results, relative presets (15m, 1h, …), All, and brush selection
• Chart aligns with Kibana Discover: absolute ISO ranges, calendar_interval, time_zone
• Results preset: histogram over current search hits without extra time filter
• Single-click bar brush applies filter immediately; dismissible filter chip
• Changing the chart timestamp field re-runs _search with that field
• Document sort follows the selected chart time field (not stale @timestamp)

Query — columns & fields

• Default visible columns from field usage stats (most-used fields first)
• Auto columns toggle (default on): refresh columns when index pattern changes
• _id no longer added automatically; add manually from Available fields if needed
• Field sidebar and column reset wait for metadata/search where appropriate

Fixes & hardening

• Index pattern switch: no spurious change on picker blur; sort sanitized per index
• Index switch no longer sends stale sort/time field (400 mapping errors)
• API gateway verification script (npm run verify:api-gateway)

Query tab

• Kibana Discover-style search bar (index pattern + query on one row)
• Index and data stream pattern picker; empty pattern defaults to all indices (*)
• Hidden indices (names starting with .) sorted to the end of the list
• Simple and advanced JSON search modes with document table, field panel, and export

Index detail

• Global index detail modal (Overview, Data, Mappings, Settings, ILM) from any index link
• Elasticvue-style document browse with drag-and-drop columns and query_string search

Import / Export

• Home page Import/Export for cluster connections (Elasticvue + EPM NDJSON backup format)

UX and branding

• SearchAli wordmark (light/dark) linking to searchali.com
• Index health dots in Index Statistics
• Dark mode readability for index/node/template names
• Shards tab: larger typography aligned with Query, sticky Nodes column
• Cluster selector and header logo sizing improvements

• ILM visibility improvements (Managed / Without ILM)
• Field usage labeling and interaction consistency updates
• Snapshot detail loading hint for long API calls
• Shards tab aborted-request UX fix
• Default API timeout increased to 30 seconds

Highlights

• Added Search tab scaffold (coming soon)
• Consolidated Index Explorer into Indices tab
• Improved search behavior with AND/NOT and diacritic-insensitive matching
• Added node allocation-backed shard/disk details in Nodes tab

Highlights

• Shards tab: ElasticVue-style Node × Index grid with shard placement, stable indexing-rate ordering, and shard detail modal.
• Shards: 10s auto-refresh, fixed request storm issues, and improved loading/empty states.
• Snapshots: Completed % + processed/total, remaining time estimate, and 10s auto-refresh with safe in-flight handling.

• Guided onboarding tour for first-time setup
• Snapshots: partial repo support + snapshot detail modal
• Normalize cluster URLs to include http(s) scheme

Release-pipeline iteration, published the same day as v1.0.0 (see above). No user-facing change.

Release-pipeline iteration, published the same day as v1.0.0 (see above). No user-facing change.

Release-pipeline iteration, published the same day as v1.0.0 (see above). No user-facing change.

Release-pipeline iteration, published the same day as v1.0.0 (see above). No user-facing change.

Release-pipeline iteration, published the same day as v1.0.0 (see above). No user-facing change.

Release-pipeline iteration, published the same day as v1.0.0 (see above). No user-facing change.

First published version — the initial package uploaded to the Chrome Web Store. v1.0.1 through v1.0.6 were released the same day, seconds apart: early iterations of the release pipeline itself, with no behavior change for users.