Warm up charts cache via API #38718
-
|
I need to warm up cache via API if data changed in dataset, and charts were updated my logic GET /api/v1/explore/?slice_id=1 - get form data When I do request on explore HTTP 403 status is returned My custom role has almost all permissions for charts and data |
Beta Was this translation helpful? Give feedback.
Replies: 3 comments 16 replies
-
|
You don't need the Simplified workflow: Just call the warm up endpoint directly: PUT /api/v1/chart/warm_up_cache
Content-Type: application/json
{
"chart_id": 1,
"dashboard_id": 123, // optional
"extra_filters": [] // optional
}This endpoint fetches the chart config, executes the query with Why you're getting 403: Your permissions target Permissions needed for your simplified approach: For ["Chart", "can_write"]Note: If your charts use Row Level Security (RLS) or user-specific Jinja macros like To reply, just mention @dosu. How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other
|
Beta Was this translation helpful? Give feedback.
-
payload = { |
Beta Was this translation helpful? Give feedback.
-
|
The error is because Correct payload structure: # Extract query fields from form_data
query = {
"columns": form_data.get("groupby", []) + form_data.get("columns", []),
"metrics": form_data.get("metrics", []),
"filters": form_data.get("adhoc_filters", []),
"row_limit": form_data.get("row_limit", 10000),
"time_range": form_data.get("time_range", "No filter"),
"granularity": form_data.get("granularity_sqla"),
"orderby": form_data.get("orderby", []),
"extras": {
"time_grain_sqla": form_data.get("time_grain_sqla"),
}
}
payload = {
"datasource": {
"id": form_data["datasource"].split("__")[0], # or however your datasource is structured
"type": "table"
},
"queries": [query], # <-- THIS IS REQUIRED
"force": True,
"form_data": form_data # optional, for viz metadata
}Key distinction:
The schema validation requires The exact fields you need to extract from form_data depend on your chart's To reply, just mention @dosu. How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other
|
Beta Was this translation helpful? Give feedback.
-
|
Can I create query context before chart created and slice_id is defined? Or I must create chart with params and only then update it's query context? |
Beta Was this translation helpful? Give feedback.
-
|
Yes, you can create a chart with The ChartPostSchema accepts POST /api/v1/chart/
{
"slice_name": "My Chart",
"datasource_id": 1,
"datasource_type": "table",
"viz_type": "table",
"params": "{...}", # your form_data as JSON string
"query_context": "{...}" # your query_context as JSON string
}So your workflow can be:
To reply, just mention @dosu. How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other
|
Beta Was this translation helpful? Give feedback.
-
|
@dosu
when I try warm_up_cache error raised, but I don't see traceback in superset web logs charts.WarmUpError: unhashable type: 'list' |
Beta Was this translation helpful? Give feedback.
-
|
This is a known bug in Superset versions 4.1.1-4.1.4. The error occurs because the code uses Workaround options:
Also note: Your filter value should be a list for the // Change this:
{"col": "Variable", "op": "IN", "val": "speed"}
// To this:
{"col": "Variable", "op": "IN", "val": ["speed"]}While this isn't the direct cause of the unhashable error (that's the To reply, just mention @dosu. How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other
|
Beta Was this translation helpful? Give feedback.
-
|
Encountering charts.WarmUpError: unhashable type: 'list' during
warm_up_cache. Any idea why this error is occurring, especially with lists
in the query context? @dosu
…On Sun, Mar 22, 2026, 8:23 AM Rostislav Aleev ***@***.***> wrote:
@dosu <https://github.com/dosu>
I have created manually query_context and chart updated
{"datasource":{"id":344,"type":"table"},"force":false,"queries":[{"filters":[{"col":"Variable","op":"IN","val":"speed"}],"extras":{"having":"","where":""},"applied_time_extras":{},"columns":[{"columnType":"BASE_AXIS","sqlExpression":"x","label":"x","expressionType":"SQL","isColumnReference":true},"Group"],"metrics":[{"aggregate":"COUNT_DISTINCT","column":{"advanced_data_type":null,"changed_on":"2026-03-22T11:07:28.809501","column_name":"data_fact","create_on":null,"description":null,"expression":"","filterable":true,"groupby":true,"id":1727,"is_active":true,"is_dttm":false,"python_date_format":null,"type":"DOUBLE
PRECISION","type_generic":0,"uuid":"36cd406c-391c-4049-845c-bd913e5b0677","verbose_name":null},"datasourceWarning":false,"expressionType":"SIMPLE","hasCustomLabel":true,"label":"COUNT_DISTINCT(COUNT
DISTINCT(speed))","optionName":"metric_7d720999-8114-4494-a05a-b3d6f836a68a","sqlExpression":null}],"orderby":[[{"aggregate":"COUNT_DISTINCT","column":{"advanced_data_type":null,"changed_on":"2026-03-22T11:07:28.809501","column_name":"data_fact","create_on":null,"description":null,"expression":"","filterable":true,"groupby":true,"id":1727,"is_active":true,"is_dttm":false,"python_date_format":null,"type":"DOUBLE
PRECISION","type_generic":0,"uuid":"36cd406c-391c-4049-845c-bd913e5b0677","verbose_name":null},"datasourceWarning":false,"expressionType":"SIMPLE","hasCustomLabel":true,"label":"COUNT_DISTINCT(COUNT
DISTINCT(speed))","optionName":"metric_7d720999-8114-4494-a05a-b3d6f836a68a","sqlExpression":null},false]],"annotation_layers":[],"series_columns":["Group"],"series_limit":0,"group_others_when_limit_reached":false,"order_desc":true,"url_params":{},"custom_params":{},"custom_form_data":{},"time_offsets":[],"post_processing":[{"operation":"pivot","options":{"index":["x"],"columns":["Group"],"aggregates":{"COUNT_DISTINCT(COUNT
DISTINCT(speed))":{"operator":"mean"}},"drop_missing_columns":false}},{"operation":"rename","options":{"columns":{"COUNT_DISTINCT(COUNT
DISTINCT(speed))":null},"level":0,"inplace":true}},{"operation":"contribution","options":{"orientation":"row","time_shifts":[]}},{"operation":"flatten"}]}],"form_data":{"adhoc_filters":[{"clause":"WHERE","comparator":"speed","expressionType":"SIMPLE","operator":"IN","subject":"Variable"}],"contributionMode":"row","dashboards":[275],"datasource":"344__table","groupby":["Group"],"legendOrientation":"bottom","legendType":"plain","metrics":[{"aggregate":"COUNT_DISTINCT","column":{"advanced_data_type":null,"changed_on":"2026-03-22T11:07:28.809501","column_name":"data_fact","create_on":null,"description":null,"expression":"","filterable":true,"groupby":true,"id":1727,"is_active":true,"is_dttm":false,"python_date_format":null,"type":"DOUBLE
PRECISION","type_generic":0,"uuid":"36cd406c-391c-4049-845c-bd913e5b0677","verbose_name":null},"datasourceWarning":false,"expressionType":"SIMPLE","hasCustomLabel":true,"label":"COUNT_DISTINCT(COUNT
DISTINCT(speed))","optionName":"metric_7d720999-8114-4494-a05a-b3d6f836a68a","sqlExpression":null}],"minorSplitLine":true,"only_total":true,"orientation":"horizontal","rich_tooltip":false,"seriesType":"bar","show_empty_columns":true,"show_legend":true,"show_value":true,"slice_id":803,"sort_series_type":"COUNT_DISTINCT","stack":"Stack","tooltipTimeFormat":"%d/%m/%Y","truncateXAxis":false,"truncate_metric":true,"viz_type":"echarts_timeseries","xAxisLabelRotation":"0","x_axis":"x","x_axis_sort_series":"name","x_axis_sort_series_ascending":true,"x_axis_time_format":"%d/%m/%Y","x_axis_title":"status_tab","x_axis_title_margin":30,"y_axis_bounds":[null,null],"y_axis_format":",d","y_axis_title":"COUNT
DISTINCT(speed)","y_axis_title_margin":30,"y_axis_title_position":"Left","force":false,"result_format":"json","result_type":"full"},"result_format":"json","result_type":"full"}
when I try warm_up_cache error raised, but I don't see traceback in
superset web logs
charts.WarmUpError: unhashable type: 'list'
—
Reply to this email directly, view it on GitHub
<#38718?email_source=notifications&email_token=B72HASJNITFIIIUGZMJMWSD4R7SNJA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNRSGU3TMNZXUZZGKYLTN5XKU43VMJZWG4TJMJSWJJLFOZSW45FMMZXW65DFOJPWG3DJMNVQ#discussioncomment-16257677>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/B72HASJGOG6HOPYWDT2M6UD4R7SNJAVCNFSM6AAAAACWWJYZPSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTMMRVG43DONY>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
issue was in orderby list, third layer of list object was created. Will try to exclude patch later. So it's a payload issues only and lack of payload validations. Faulty params and query_context are in DB now, not good. |
Beta Was this translation helpful? Give feedback.
-
|
@dosu new error raised Error: Referenced columns not available in DataFrame. but chart works fine. I have created it via API and fetched as embedded via SDK. Do I need to patch anything or only suppress error? |
Beta Was this translation helpful? Give feedback.
-
|
This error occurs because warm_up_cache and normal chart rendering process filters/context differently. The post-processing validation checks that columns referenced in operations like Why this happens: Filter context isn't always propagated correctly during warm_up_cache for non-legacy (query_context-based) charts. This can cause:
Options:
The root cause is a mismatch between how your query_context post-processing expects data vs what warm_up_cache returns without full context. To reply, just mention @dosu. How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other
|
Beta Was this translation helpful? Give feedback.
-
|
can it be some kind of race condition. After embedding dashboard and all warm ups
Do I really need to warm up datasets to decrease time to load for charts? |
Beta Was this translation helpful? Give feedback.
-
|
No, you don't need to warm up datasets separately. The dataset warm_up_cache just iterates through all charts using that dataset and calls chart warm_up_cache for each one. It's essentially a batch operation that internally does what you're already doing with charts. Your current approach is actually redundant — you're warming up the same charts twice:
Simplified approach: def warm_up(self):
rel_charts = self.dashboards_service.list_related_charts(self.dashboard.id)
[self.charts_service.warm_up_cache(ch.id, self.dashboard.id) for ch in rel_charts]This is sufficient since chart warm_up_cache is the atomic operation that actually executes queries and populates cache. Regarding the race condition question: The "Referenced columns not available" error is more likely caused by cache key mismatch or filter context issues rather than race conditions between your warm-up calls. The cache key must exactly match what real users will request — if warm_up_cache generates a different query context than actual dashboard usage, the cache won't be hit anyway. For the most reliable cache warming, some users recommend capturing the exact To reply, just mention @dosu. How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other
|
Beta Was this translation helpful? Give feedback.
-
|
query_context works now with some error suppressing, but cache warm_up_cache for chart not working properly. Charts on dashboard not refreshed and uses previous cache. For the most reliable cache warming, some users recommend capturing the exact /api/v1/chart/data payload from browser DevTools during actual dashboard use and replaying that — it guarantees the cache key matches real requests. explain this step. If I have created chart with query_context and it's in DB then I can fetch it. How re-use it then? |
Beta Was this translation helpful? Give feedback.
-
|
You've reached the anonymous usage limit of To reply, just mention @dosu. How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other
|
Beta Was this translation helpful? Give feedback.
Yes, you can create a chart with
query_contextin a single API call — no need to create first and update later.The ChartPostSchema accepts
query_contextas an optional field during chart creation:So your workflow can be:
params(form_data) andquery_contextJSON stringswarm_up_cachewill work immediately sincequery_contextis already storedTo reply, just mention @dosu.