## Problem
The per-service logs pages (e.g. `/project/[ref]/logs/auth-logs`)
crashed with `RangeError: Invalid time value` (Sentry issue 7580074952).
`calcChartStart` guarded `iso_timestamp_start` only against falsy
values, so a truthy-but-unparseable timestamp (a malformed value in the
URL query params) produced an Invalid Date, which propagated through
`.add()` and threw when `startOffset.toISOString()` was called.
The bug is on the legacy (non-OTEL) chart query path. The OTEL bucket
helper had the same unguarded pattern; it did not crash but could skew
the chart bucket size.
## Fix
Validate parsed timestamps with `dayjs().isValid()` and fall back to
now, matching the existing empty-param behavior. Applied to both
`calcChartStart` (legacy) and `otelChartTruncFn` (OTEL).
- Valid params produce identical output (existing tests unaffected)
- Empty params still fall back to now
- Malformed input no longer throws
Added regression tests to `Logs.utils.test.ts` and
`Logs.utils.otel.test.ts`.
## How to test
- Run the logs unit tests: `pnpm test:studio` (or target
`Logs.utils.test.ts` and `Logs.utils.otel.test.ts`)
- In the dashboard, open a service logs page with a malformed timestamp
in the URL, e.g. `/project/<ref>/logs/auth-logs?its=not-a-date`
- Expected result: the page renders without crashing and the chart falls
back to the default (now-based) time range
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Bug Fixes**
* Improved log chart time-range handling by safely resolving missing or
invalid ISO timestamps via a shared timestamp resolver.
* Updated chart bucketing and timestamp conflict logic to use the
resolved endpoints, preventing errors and ensuring correct fallback
granularity (including minute-level bucketing when needed).
* **Tests**
* Added regression coverage to confirm chart query generation (including
OTEL queries) does not throw for unparseable start/end timestamps.
* Verified fallback behavior to minute-level bucketing and non-throwing
behavior for timestamp conflict handling.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
## What
Enforces a `LIMIT` clause on Logs Explorer queries, replacing the
previous soft warning.
- Queries without a `LIMIT <n>` can no longer be run.
- Instead of the warning badge, the results box shows a clear "Add a
LIMIT to your query" message.
- The error clears as soon as a valid `LIMIT` is added.
## Why
Unbounded queries can scan very large amounts of data. This adds a UI
guardrail so a bounded result set is always requested.
## Notes
- New `checkForLimitClause` util detects `LIMIT <n>` outside of string
literals and comments (mirrors the existing WITH/ILIKE checks), with
unit tests.
- The missing-limit message reuses the existing error-rendering path via
a `missingLimit` reason, alongside `resourcesExceeded`.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Log queries now require a `LIMIT` clause before they can run.
* A new on-screen message guides users to add a `LIMIT` when it’s
missing.
* **Bug Fixes**
* Improved log query validation to better detect valid `LIMIT` usage,
including mixed case, multiline queries, and avoidance of false matches
in comments, strings, or column names.
* Existing error messages continue to appear for other query limits and
failures.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
## Problem
High availability (Multigres) projects don't expose Multigres service
logs in the Studio logs UI, so users on HA projects have no entry point
to inspect them.
## Fix
Add a `Multigres` logs collection, gated behind the `multigresLogs`
ConfigCat flag **and** the project's `high_availability` flag
(`useShowMultigresLogs`):
- New `Multigres` entry in the logs sidebar `Collections`, linking to a
new `multigres-logs` page that queries the `multigres_logs` table.
- Wire `multigres_logs` through the logs constants, types, table SQL,
query type, and service labels.
- Row formatting: parse the JSON `event_message` and render `level`
through `SeverityFormatter` and `msg` through `TextFormatter`, matching
the other service collections (instead of dumping raw JSON).
- `WARN` severity is now styled like `WARNING` (amber), since Multigres
emits `level: WARN`.
- Log detail drawer: parse the JSON `event_message` and spread its keys
onto the log so each field (level, msg, query, error, connection_id,
etc.) renders as its own collapsible row.
- Single-log query omits the `metadata` column for `multigres_logs` (the
table has no such column), fixing an `INVALID_ARGUMENT` error when
opening the detail drawer.
- Event chart: parse the level out of `event_message` via `JSON_VALUE`
so error/warning bars are counted (the table has no top-level level
column).
- Add the `multigres_logs` source schema to the Field Reference drawer,
same gating.
## Why a feature flag
`high_availability` is an existing product feature that predates
Multigres, so existing HA projects could otherwise see a broken
collection querying a `multigres_logs` table they don't have. Requiring
the `multigresLogs` flag ships the feature dark and decouples rollout
from HA status. The flag must be created in ConfigCat before enabling;
until then `useFlag` returns false and the feature stays hidden.
## How to test
- Enable the `multigresLogs` flag (or override locally) and open a
project where `high_availability` is `true`.
- Navigate to `Logs`. Confirm a `Multigres` entry appears under
`Collections` (after `Replication`).
- Open it: the page loads at `/project/<ref>/logs/multigres-logs` and
queries `multigres_logs`.
- Confirm rows show a colored severity pill (including amber `WARN`) and
a readable message rather than raw JSON.
- Confirm the chart counts error/warning bars correctly.
- Click a row: the detail drawer shows each parsed field as its own row,
with no error.
- Open the Field Reference drawer and confirm `Multigres` is listed as a
source.
- With the flag off, or on a non-HA project, confirm the collection and
Field Reference source are both hidden.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Multigres added as a dedicated log source with its own Logs page,
sidebar entry, and query type.
* Log list and preview now parse Multigres payloads to surface
timestamp, severity, and formatted message.
* Multigres integrated into charting, prompt labels, and field-reference
UI (hidden unless enabled).
* New hook controls showing Multigres UI only when feature flag + HA
project condition are met.
* **Bug Fixes**
* Severity rendering treats "WARN" the same as "WARNING".
* **Tests**
* Unit tests added for Multigres parsing and the show-Multigres hook.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Closes
[FE-3378](https://linear.app/supabase/issue/FE-3378/featlogs-keyboard-shortcuts-for-function-logs-invocations-and-logs).
## Summary
Adds a shared shortcut registry for every `LogsPreviewer` surface —
Function Logs, Function Invocations, and the Logs Explorer — and brings
the grid keyboard model in line with the Auth Users / Table Editor
patterns.
## Shortcuts
| Key | Action |
| --- | --- |
| `↑` / `↓` | Move single-row selection; opens side panel |
| `Shift+Space` | Toggle current row in multi-select |
| `Mod+A` | Toggle all visible rows in multi-select |
| `Esc` | Staged: clear multi-select → close side panel |
| `Shift+R` | Refresh logs |
| `Shift+H` | Toggle histogram |
| `Shift+L` | Load older logs |
| `Shift+P` | Open time range picker |
| `Mod+Shift+J / M / C` | Copy selected rows as JSON / Markdown / CSV
(existing global handler) |
## Other changes
- `ShortcutTooltip` on search, refresh, histogram, load older, and
time-picker controls.
- `onSearchInputEscape` wired on the logs search bar (clear → blur).
- Visual row highlight (`rdg-row--focused`) when a row is
keyboard-focused or multi-selected.
- Multi-select copy dropdown gains a **Copy as CSV** entry and shows the
keybind on each item via `ShortcutBadge`.
- Manual arrow-nav (`navigate()`) updates `selectedRow` directly without
going through `onRowClick`, so multi-select checkmarks survive keyboard
navigation.
## Test plan
- [x] Function Logs and Function Invocations: all shortcuts above fire
while the page is mounted, no firing in other tabs.
- [x] Logs Explorer: same shortcuts work; copy keybinds still copy *all*
rows when nothing is multi-selected.
- [x] Arrow keys on first load select the first row even when the focus
sink is the active element.
- [x] Selecting rows via checkbox or `Shift+Space`, then pressing arrow
keys, preserves the checkmarks.
- [x] Escape on a populated search input clears it; Escape on an empty
input blurs it.
- [x] Esc with multi-select active clears the selection before closing
the side panel.
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* CSV export for log selections (adds CSV alongside JSON and Markdown).
* New logs-preview keyboard shortcuts: search focus, refresh, chart
toggle, date picker, load older, navigation, and selection.
* **Improvements**
* Shortcut badges and tooltip integration across the logs UI.
* Search input focus/ref support and controlled date-picker visibility.
* Better no-results/error rendering and expanded copy dropdown sizing.
* **Tests**
* Added CSV formatting tests covering RFC 4180 edge cases.
<!-- review_stack_entry_start -->
[](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/45989)
<!-- review_stack_entry_end -->
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## Problem
The edge function logs table at
`/observability/logs/edge-functions-logs` did not show which edge
function was called. Users had to open individual log entries to find
the function name, even though the information is available in the log
data.
## Fix
- Added `request.pathname` to the SQL SELECT query for
`function_edge_logs` so the pathname is returned with each row.
- Added `extractEdgeFunctionName` utility that parses the last path
segment from the pathname (e.g. `/functions/v1/hello-world-1` becomes
`hello-world-1`).
- Updated `FunctionsEdgeColumnRender` to display the function name
between the HTTP method and the log ID columns.
- Added unit tests for `extractEdgeFunctionName` covering normal paths,
null/undefined, non-string values, and pathnames without slashes.
<img width="1472" height="776" alt="CleanShot 2026-03-19 at 13 18 47@2x"
src="https://github.com/user-attachments/assets/e437f7df-7aab-4ea5-b421-95bab3923605"
/>
## How to test
- Navigate to Logs and Analytics, then Edge Functions logs.
- Confirm each row now shows the function name (e.g. `hello-world-1`)
between the HTTP method and the log ID.
- Confirm rows without a pathname (e.g. non-platform or self-hosted)
still render without errors.
- Confirm clicking a row still opens the log detail panel correctly.
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Allows users to select logs and copy their contents for debugging, or
pass them on to assistant with one click.
## To test
- go to logs or log explorer
- select some logs
- try copying as json, markdown, or sending it to assistant.
