Dashboard page stuck on loading after upgrading superset to 6.1.0

[Dhruvin4530]

Dashboard page stuck on loading after upgrading superset to 6.1.0 & then black out

@dosu Please take a look.

[dosubot[bot]]

Hi @Dhruvin4530, this is a commonly reported issue after upgrading Superset. Based on similar reports, here are the most likely causes and troubleshooting steps:

1. Stale browser cache (most common cause)

After an upgrade, your browser may still be referencing old JavaScript/CSS chunks from the previous version. Try:

• Hard refresh: Ctrl+Shift+R (Windows/Linux) or Cmd+Shift+R (Mac)
• Clear your browser cache entirely

2. Node version mismatch during frontend build

Superset 6.1.0 requires Node ^20.16.0 (recommended v20.18.3). If your frontend assets were built with an older Node version, they may be broken [1]. Rebuild with:

npm ci && npm run build

3. Dependency version conflicts

If upgrading via Docker, stale volumes can leave outdated dependencies (e.g., sqlglot v27.x instead of the required version). Clear all Docker volumes and rebuild fresh [2]:

docker compose down -v
docker compose up --build

4. Theme configuration issue

If you have a partial THEME_DEFAULT override in your config, it may cause a blank/black screen. A fix was merged in PR #41347 that deep-merges partial theme overrides with built-in defaults [3].

5. Dashboard scrolling regression

There was a known regression in 6.x where dashboards taller than the browser window were clipped with no scrollbar (appearing as a blank/black area). This was fixed in PR #41439 [4].

Recommended troubleshooting order:

1. Hard refresh / clear browser cache
2. Check browser developer console (F12 → Console tab) for specific errors like ChunkLoadError or TypeError
3. Verify Node version if you built frontend yourself
4. Run superset db upgrade to ensure migrations are complete
5. If using Docker, rebuild with clean volumes

Could you check your browser console for any error messages? That would help narrow down the exact cause.

To reply, just mention @dosu.

---

Docs are dead. Just use Dosu.

[dosubot[bot]]

Thanks for sharing your config, @Dhruvin4530. I can see a likely culprit — your THEME_DEFAULT partial override is almost certainly causing the black screen.

The problem: Your config defines:

THEME_DEFAULT = {
    "token": {
        "brandAppName": APP_NAME,
        "brandLogoAlt": APP_NAME,
    },
}

In Superset 6.x, when you override THEME_DEFAULT with only a few token fields, it completely replaces the built-in theme dictionary (which contains ~22 required token fields like fontFamily, colorPrimary, colorLink, etc.). The frontend then crashes trying to access these missing fields, resulting in the blank/black dashboard you're seeing [1].

The fix: PR #41347 (merged June 24, 2026) resolves this by deep-merging partial overrides with the built-in defaults [1]. However, depending on when you pulled the apache/superset:6.1.0 image, this fix may not be included yet.

Immediate workaround — remove or comment out your theme overrides:

# THEME_DEFAULT = {
#     "token": {
#         "brandAppName": APP_NAME,
#         "brandLogoAlt": APP_NAME,
#     },
# }

# THEME_DARK = {
#     "token": {
#         "brandAppName": APP_NAME,
#         "brandLogoAlt": APP_NAME,
#     },
#     "algorithm": "dark",
# }

If you need custom branding, you'll need to either:

1. Wait for an image that includes PR fix(theming): deep-merge partial THEME_DEFAULT overrides with built-in defaults #41347's fix
2. Or provide the full theme dictionary with all required fields (you can copy from Superset's built-in config.py and just change the fields you care about)

Additionally, there's a separate dashboard scrolling regression from @visx/responsive v4.0.0 that clips dashboards taller than the viewport, which was fixed in PR #41439 [2]. This could also contribute to the appearance of a blank page if your dashboard content is below the fold.

Try removing the THEME_DEFAULT and THEME_DARK overrides first — that's the most likely cause given what you're seeing.

To reply, just mention @dosu.

---

Docs are dead. Just use Dosu.

[Dhruvin4530]

@dosu Only 1 dashboard is failing to load, all other dashboards are loading fine. The dashboard is getting data from hydrolix and it is running so many queries.

[dosubot[bot]]

Thanks for the additional context — if only one dashboard is failing and it's the one querying Hydrolix with many charts, the THEME_DEFAULT issue from my earlier suggestion is likely not the cause.

Since you're using Hydrolix via ClickHouse and upgraded to 6.1.0, there's a relevant compatibility issue: your Dockerfile installs clickhouse-connect>=0.13.0, and this newer driver version appends FORMAT Native to queries sent over HTTP. When combined with Hydrolix-specific SETTINGS (like hdx_query_timerange_required), this can cause Hydrolix's time-range validator to reject queries — even when valid time filters are present [1]. If every chart on that dashboard hits this validation failure, the entire dashboard would get stuck loading and eventually black out.

To confirm this theory:

1. Open the browser dev console (F12 → Network tab) and look for failed API calls to /api/v1/chart/data returning errors
2. Check Superset logs for errors like "hdx_query_timerange_required is set to true. Your query needs a time range filter" or similar ClickHouse/Hydrolix-related failures

Possible fixes:

• If you have a SQL_QUERY_MUTATOR adding Hydrolix settings, try conditionally disabling hdx_query_timerange_required or adjusting the mutator logic to account for the new query format
• Alternatively, try pinning clickhouse-connect to a pre-0.13.0 version temporarily to see if that resolves it:

  RUN uv pip install "clickhouse-connect>=0.7.0,<0.13.0"

If that's not the issue, the problem could also be a query timeout/concurrency issue — a dashboard running many heavy Hydrolix queries simultaneously may be overwhelming the connection or hitting timeout limits. In that case, check whether individual charts from that dashboard load fine in Explore view.

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

[Dhruvin4530]

@dosu

This is my query mutator script.

def SQL_QUERY_MUTATOR(sql, **kwargs):
    """Query mutator compatible with Superset v5.0+.

    In Superset v2, the signature was:
        SQL_QUERY_MUTATOR(sql, security_manager, database, **kwargs)

    Since Superset v2.0 (PR #19083), the signature changed to:
        SQL_QUERY_MUTATOR(sql: str, **kwargs: Any) -> str

    where security_manager, database, user_name, etc. are passed via kwargs.
    """
    # Extract arguments from kwargs (v5 passes everything via kwargs)
    security_manager = kwargs.get("security_manager")
    database = kwargs.get("database")

    # Sometimes we get a different object here.
    # But it should at least be convertible to a string that represents the actual query.
    log.debug(f"SQL_QUERY_MUTATOR")
    log.debug(f"kwargs: {kwargs}")
    query = str(sql)

    # Don't mutate if this isn't a SELECT query
    select_search = re.search(r"^\s*(SELECT|WITH)", query, flags=re.IGNORECASE)
    if not select_search:
        return query

    # Extract any SETTINGS already set in the query
    query, query_settings = extract_settings(query)

    # Get username - v5: get_username() was removed, use Flask g.user instead
    user_name = _get_username()

    # Determine the SETTINGS for this query
    new_settings = get_settings(
        user_name=user_name,
        security_manager=security_manager,
        query_settings=query_settings,
    )
    database_name = database.database_name
    log.debug(f"database_name: {database_name}")
    # Add the SETTINGS to the query and return it.
    if "hydrolix" in database_name.lower():
        # Fix series limits to use LIMIT BY instead of subquery.
        query = replace_series_limit(query)
        query = rewrite_array_join_filter_conditions(query)

        query = hdx_with_settings(query, new_settings)
    if "athena" in database_name.lower():
        query = athena_with_settings(query, new_settings)
    return query
    ```

[dosubot[bot]]

You've reached the anonymous usage limit of 10 interactions per month. Your usage will reset on 2026-07-01. If you'd like to use Dosu before then, please register to increase your limit to 20 interactions per month for free.

To reply, just mention @dosu.

---

Docs are dead. Just use Dosu.