## Stabilize the Alpha1 (cloud) e2e suite + regression carryover
Root-caused and fixed the failures surfacing on the Alpha1 cloud runs —
investigated live via Playwright MCP against alpha and, where possible,
a local build. Grouped by theme. Every fix is non-masking (still fails
if the underlying is genuinely broken). Pairs with ENT #2351
(`max-parallel` + Logs-SelectStar `quick_mode`).
### Runtime / infra resilience (whole-suite)
- **Auth self-heal on 401 / session expiry** — the shared cloud session
+ per-org passcode are minted once and reused all run; on
load/contention they expire mid-run (`User not authenticated`, `HTTP 401
Unauthorized`). `cloud-auth.authedRequest` refreshes the passcode and
retries on 401 (escalating to a full Dex re-login if the session is
dead); `navigateToBase` re-authenticates and resumes instead of
throwing. Also routed the alert **template-creation** API through it
(401 there cascaded into "Failed to navigate to templates"). Gated on
`isCloudEnvironment()` — no self-hosted change.
- **`--disable-dev-shm-usage`** on the alpha1 Chromium launch — the EKS
pods used the default 64MB `/dev/shm`; under 5 workers on long/heavy
shards Chromium exhausted it and the pod OOM-killed → "runner lost
communication" (frozen step + 404 logs). Route shared memory to /tmp;
eliminates the infra deaths.
- **Service graph queries a wide window** (`period=6h`) — the tab
defaulted to Past 15 Minutes and rendered "No service graph data" before
the daemon-processed topology fell inside the window.
### Pipelines
- `queryStream(stream, expectedMinCount)` **polls until the record count
lands** (pipeline output arrives async) — was single-shot ("Expected 5,
got 3").
- `openPipelineForEdit` **searches + polls-with-reload** for the edit
button (list not refreshed after save → 45s click timeout).
- `exploreStreamAndNavigateToPipeline` **reloads the streams list**
while polling for Explore, and **falls back to a direct logs URL** when
a freshly-ingested source stream never surfaces in the list
(stream-stats lag).
### Alerts
- Alert-destination URL → **public `example.com` webhook** (self-host
URL hit the cloud **SSRF guard** → form never closed).
- **Force-click the virtualized folder item** (#9484) — the row
re-rendered mid-click → 45s "not stable".
- **Export button readiness** — it's rendered only on selection; wait
for visible+enabled (re-selecting if needed) before clicking.
- **`deleteAlertByRow` cleanup readiness** — the just-created alert's
more-options kebab needs longer than 5s under load.
### RegressionSet
- **Use the shard's org** (`getOrgIdentifier()`), not hardcoded
`org_identifier=default`, in `logs-bugs` — on cloud the data lives in
`auto_org_N`, so the logs page never rendered.
- **Unique stream names** for `logs-9754` / `logs-bugs:1419` — fixed
names collided with concurrent shared-org runs ("stream is being
deleted").
### Traces / correlation
- **Migrated `logs-bugs:995` (bug-11469)** to the redesigned correlation
UI — the old `log-correlation-btn` / `correlation-dashboard-close` are
gone; repointed to the inline `correlated-traces-tab`, assertion
(`expectNoContextMenuVisible`) unchanged and **verified live on alpha**
(no fake-pass).
### Workflows (ENT-only) + misc
- **Healed the workflows destination flow** and forced-type
Connection-step assertion; alert/report list-readiness hardening.
- **Enrichment tables list** readiness (30s + re-nav retry) — "slow
under load".
### Notes
- Validated live via MCP (alpha) and a local build where the feature
exists.
- Alpha1 config already runs `retries: 2`; the remaining first-attempt
flakes are shared-env load, addressed on the ENT side via
`max-parallel`.
## What changed
Naming an alert, panel, workflow, pipeline or function meant inventing a
name up front in a form field before anything else could be configured.
Two pieces replace that.
**Inline rename** — the page title *is* the input. `OInlineEdit` /
`OFormInlineEdit` (`web/src/lib/forms/InlineEdit`): click to rename,
Enter or blur commits, Escape restores the pre-edit value. A `#trail`
slot carries the "Auto" badge in display mode only.
**Auto-generated names** — `utils/autoName.ts` builds the name,
`composables/useAutoName.ts` decides when it's allowed to win. A name is
AUTO until the user types, then it's theirs for good; clearing the field
and blurring re-arms auto. The generators describe only what is actually
configured, so a half-built panel yields a shorter name rather than a
speculative one, and an empty string when there is nothing to say.
- Panel names read as prose — `Avg of duration by service`, `Record
count`, `k8s_logs overview`. Both live panel-field shapes are handled
(builder `functionName` + `args`, and legacy flat `column` +
`aggregationFunction`); the time column is read from
`zoConfig.timestamp_column` rather than hardcoded. Hand-written queries
fall back to the stream, since naming a panel after `x_axis_1` is worse
than saying nothing.
- Alert names are identifiers — `anomaly_k8s_logs`,
`k8s_logs_status_gte_500` — built to survive the backend's
`RE_OFGA_UNSUPPORTED_NAME` rule, with operators spelled out and a
`slugify` backstop. A leading underscore is preserved so internal
streams (`_rumdata`) keep their names.
### Supporting library changes
| Component | Change |
| --- | --- |
| `OPageHeader` / `OPageLayout` | New `titleOverflow="visible"` — the
`<h1>`'s `truncate` clips an interactive title's focus ring and the
error message it floats below itself |
| `OSelect` | New `appearance="inline"` — borderless, no fixed height,
sized to content, for a value that reads as a word inside running text
("Add Alert **in KTX**") |
All new strings go through i18n (`common.inlineEdit.*`,
`dashboard.autoName.*`, `alerts.autoName.*`, per-module `renameHint`).
Interpolated identifiers — column names, SQL functions, stream names —
stay verbatim.
## Wired into
AddPanel, AddAlert (+ `useAlertForm`), WorkflowEditor, PipelineEditor,
Functions / FunctionsToolbar, IncidentDetailDrawer,
InlineSelectFolderDropdown.
## Tested
244 unit tests across the 5 affected suites, all passing — including new
specs for `OInlineEdit`, `useAutoName`, and `autoName`.
One pre-existing expectation in `autoName.spec.ts` was updated: it
asserted `Avg of duration, max of duration +1 more`, capitalising only
the first word of a multi-measure name. `measureLabel` capitalises per
measure by design, so the expectation now reads `Avg of duration, Max of
duration +1 more`.
---------
Co-authored-by: ktx-vaidehi <134508096+ktx-vaidehi@users.noreply.github.com>
Co-authored-by: ktx-vaidehi <vaidehi.akhani@kiara.tech>
## How this was found
The **Playwright E2E Crosscheck** bot (the ENT gate that runs
enterprise-only specs against OSS PRs) flagged a failing spec on an
unrelated logs PR ([#13414
comment](https://github.com/openobserve/openobserve/pull/13414#issuecomment-5058151842)):
> `Workflows/workflows-negative.spec.js` — *"enterprise-only e2e spec
failed against this PR … wouldn't show up in OSS CI."*
Digging showed it wasn't caused by that PR — the same test was **also
failing on #13413 and #13315**. It's a pre-existing **flake** in the ENT
gate, so this PR fixes it at the source.
## Evidence it's a flake, not a real break
The **same branch** produced both outcomes:
| ENT crosscheck run | PR | K9 test |
|---|---|---|
| 30005431754 | #13414 | ✅ passed (~50s, slow path) |
| 30008606974 | #13414 | ❌ failed |
| 30006996158 | #13315 | ❌ failed |
| 30006419247 | #13413 | ❌ failed |
## Root cause
`CT-19 K9` hard-asserted that a normal `.click()` on Save is intercepted
by a tooltip overlay:
```js
expect(await pm.workflowsPage.normalSaveClickIsIntercepted(4000)).toBe(true);
```
But the Save button has **no tooltip of its own** (`WorkflowEditor.vue`)
— the overlay that intercepts it is a **transient/adjacent tooltip**
that often never appears on a programmatic hover in headless CI. The
probe therefore conflates *"blocked by overlay"* with *"click landed"*,
and the assertion flips run to run. (A first attempt — waiting for the
overlay before probing — didn't help: the overlay simply isn't reliably
summoned.)
## Fix (test-only, no product/Vue changes)
- Treat interception as an **informational canary** —
`testLogger.info(...)` instead of a hard assert.
- Keep the **stable** assertion: the JS-click Save workaround still
reaches the app (`msg.length > 0`) — the adjacent trigger-only-save test
already exercises that same path and passes consistently.
- If K9 is ever genuinely fixed, the canary log shows
`intercepted:false` consistently — the signal to drop the JS-click
workaround.
This removes the flaky gate without losing meaningful coverage. Changes
are limited to the spec and its page object.
## Verification
`Workflows/*` are enterprise-only (skipped on OSS CI), so verified via
the **ENT gate against this OSS branch** (`e2e` label). The removed
assertion was the only flaky part; the retained assertion matches an
already-stable adjacent test.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Ashish KolheandVamsi Karthik R
BhargavandNeha P
