Apache Superset Support, Upgrades and Managed Operations
Last reviewed by Tyler Eastridge
AceMQ gives data teams Apache Superset engineers on call 24/7 and for planned work: upgrades that run the metadata database migrations safely, a production layout for the web servers, Celery workers and Redis, and SSO with roles and row-level security that survive an upgrade. The Apache project supplies documentation and community help, and Preset sells a hosted service. AceMQ supports the Superset you host yourself, and can run it for you on your infrastructure or ours.
A senior engineer, not an SDR. No sales sequence attached.
Without a dashboard outage to find out what the upgrade changedFor production incident cover, see AceMQ Apache Superset 24/7 support.
Customers
Partners
The Problem
Eight Apache Superset problems that teams bring to AceMQ, from a quickstart database in production to a second beat sending every report twice and permissions that vanish after an upgrade. If one describes your week, that is where the review starts.
Production still runs on the SQLite database from the quickstart, or on a container volume nobody has backed up or restored
An upgrade ran superset db upgrade against the live metadata database with no backup, and UPDATING.md warns that some migrations cannot be undone cleanly
Scheduled reports arrive twice because a second Celery beat instance was started, and the docs say to run only one
Async queries in SQL Lab hang because the workers and the web server do not share the same Celery configuration or results backend
Dashboards that loaded in seconds now hit the data warehouse on every view, because the cache layer was lost or its timeouts are too short
SSO works for the first administrator, then new users land with the wrong role because role mapping and the registration role were left at their defaults
Permissions edited by hand on the Public role are overwritten after an upgrade, because superset init syncs that role when PUBLIC_ROLE_LIKE is set
Nobody has written down how to rotate SECRET_KEY, which needs PREVIOUS_SECRET_KEY and a re-encryption of stored secrets
A good fit if you self-host Apache Superset in production and need 24/7 cover, an upgrade you can rehearse, working SSO and roles, or a written view of the metadata database and workers before the next release.
Not a fit for small teams with non-critical workloads. AceMQ is not a budget consultancy: we work with enterprises running business-critical and compliance-heavy Apache Superset.
How it breaks, and the fix
Superset's cache layer does two jobs: it stores query results so a chart loaded twice is served from the cache, and it acts as the message broker for the workers. Where one Redis does both, losing it hits the warehouse and the workers at once.
Redis lost, charts query the warehouse directly
Viewers reach Superset web, which uses a warehouse, a Redis cache and broker, and Celery workers
Charts are served from the cache after their first load, so the warehouse sees each query once. The workers take async SQL Lab queries, scheduled reports and thumbnails from the broker, and the metadata database holds the dashboard definitions.
The cache is gone, so each chart view runs its query on the warehouse and load climbs with the number of viewers. The workers lose their broker, so async queries, alerts and reports and thumbnails stop being picked up.
Redis is back or has failed over. The first views refill the cache, warehouse load falls, and the workers reconnect or are restarted and take up scheduled jobs again. Anything queued only in the lost broker has to be rerun.
The loss is the easy part to see. What decides the outcome is set earlier: whether the cache and the broker are separate Redis instances, whether Redis is replicated or backed by a managed service, how long cached results live, whether async queries and alerts are in use at all, and whether anyone is alerted when the broker disappears and not only when a dashboard errors. The Superset documentation calls the caching layer optional but necessary for some features, so a layout that skips it can look fine until the first scheduled report is missed.
Our Services
AceMQ runs five Apache Superset service lines: upgrade and migration planning, production architecture and high availability, SSO with roles and row-level security, performance and warehouse load review, and a fixed-scope health check. Any of them can roll into 24/7 support with a 15-minute emergency SLA.
A route from the version you run to the one you want, built from the project's UPDATING.md and tried on a copy of your metadata database first.
A layout that survives a node loss: more than one web process, Celery workers, a single beat, Redis and a metadata database that is managed and backed up.
Sign-in through your identity provider, roles that map from its groups, and row-level filters that hold for dashboards and, where enabled, SQL Lab.
Why a dashboard is slow, read from where the time goes: the query on your warehouse, the cache, or the application.
The fixed-scope entry engagement: a written report on the state of the installation that your own team can act on, with or without AceMQ afterwards.
Senior engineers on your account, a 15-minute emergency response, no ticket routing and no junior triage.
24/7 Support
AceMQ Apache Superset support gives you an AceMQ engineer and a 15-minute first response when dashboards are down or failing, 24/7. Lower severities get 1-hour, 4-hour and next-day responses. Fixes still come from Apache Superset releases: AceMQ does not ship its own builds.
P1 escalation path
A replay of one P1, from the first page to the written root-cause analysis
15-minute P1 response, 24/7.
A response time, not a resolution time.
P1 opened, on-call engineer paged
Phone, email or Slack reaches the on-call senior engineer directly, whether the incident is in RabbitMQ, Kafka, Redis, PostgreSQL or Kubernetes. No web form, no tier-1 queue.
Senior engineer on your bridge
A senior engineer who already knows your environment joins a live bridge, so nobody has to re-explain your topology.
Specialists across the stack join
When an incident crosses components, the engineers who know each one join. For RabbitMQ, the path runs to the core team.
Written root-cause analysis
Root cause, the fix applied and the prevention steps, delivered after every P1.
Beyond SLAs — the model behind senior-only, zero-cold-start expert access.
Tickets go to senior engineers briefed on your environment, not an anonymous queue. The proposal states who does the work.
Any ticket can be escalated to a live session with an engineer through calendar booking. No gatekeeping and no approval step.
Scheduled health checks on your deployment and alerts on risks in the versions you run, so problems surface before they reach production.
Proactive alerts on critical bugs and CVEs affecting your exact version, with version compliance monitoring so you're never caught off guard.
How It Works
An AceMQ Superset engagement runs in four stages: a scoping call with a quote within 1 hour, a read-only review, a plan rehearsed on a copy of the metadata database, then delivery with your engineers and a handover. Each stage ends in writing.
Quote within 1 hour
A senior AceMQ engineer asks what you run: Superset version and install method, the metadata database, Redis and Celery layout, how many users and dashboards, how sign-in works and who is on call today. A quote request is answered within 1 hour.
Scoped per installation
A written review of version position, the metadata database and its backups, cache and worker settings, authentication, roles and row-level security filters. Findings are ranked by production risk before anything changes.
Scoped per installation
An upgrade or hardening plan tried on a copy of the metadata database and your configuration, with the order of steps, the backup point and the way back written down. Nothing is rehearsed on the live system.
Scoped per installation
Changes made with your engineers in an agreed window, then support or operations under the contract you choose. You keep the configuration, the runbooks and the written results.
A senior engineer, not an SDR. No sales sequence attached.
Typical engagements
Typical Apache Superset engagements from AceMQ: what the environment looks like, what AceMQ assesses or changes, and what the engagement delivers. Each card links to the full write-up.
Analytics Platform Team
Upgrading a self-hosted Apache Superset across releases with the metadata database migrations rehearsed, and putting sign-in, roles and row-level security on the company's identity provider.
PostgreSQL
Redis
Global Payments Processor
Emergency remediation of an Elasticsearch cluster that would not return to green after a data node failure, with replicas blocked by disk watermarks.
Elasticsearch
Kibana
Linux
KubernetesSaaS Observability Vendor
Ongoing 24/7 support for an Elasticsearch estate suffering repeated parent circuit breaker trips and long garbage collection pauses under aggregation load.
ElasticsearchKibanaKubernetes
JavaSolutions
AceMQ covers Apache Superset end to end. Projects: consulting and implementation, health checks and migrations. Ongoing cover in four tiers, from 24/7 support and incident management to proactive monitoring, managed Apache Superset on your infrastructure and fully managed Apache Superset on AceMQ infrastructure, plus end-of-life versions.
Projects
Fixed-scope work led by a senior engineer: architecture and new builds, a health check of what you run today, or a migration with a reversible cutover.
Support and managed services
Four tiers of ongoing cover once you are live. Each tier hands more of the day-to-day work to AceMQ, from a 15-minute emergency SLA to AceMQ hosting and running Apache Superset for you.
Compare the four Apache Superset support tiersSide by side
All three run Apache Superset. The differences are who operates it, who sets the upgrade date, where your data and sign-in live and who answers at night. Each row below answers one question about the choice.
| Question | Run it yourself | Host it with Preset | AceMQ support or managed |
|---|---|---|---|
| Who operates it? | Your team runs the web servers, Celery workers, Redis and the metadata database. | Preset runs the hosted service. Its Enterprise plan lists a Managed Private Cloud option. | Your team with AceMQ on call (support), or AceMQ engineers on your infrastructure (Tier 3) or ours (Tier 4). |
| Who sets the upgrade date? | You do. The project documents the commands (superset db upgrade, then superset init) and lists breaking changes per release in UPDATING.md. | Preset, since it runs the service. Check its terms for how and when versions change. | You do, with AceMQ rehearsing the upgrade on a copy and writing the way back down first. |
| Where do data and sign-in live? | Inside your network, with your identity provider. | In Preset's service by default. Private cloud options depend on the plan, so read its pricing page. | Inside your network at support and Tier 3. On AceMQ infrastructure at Tier 4, with data handling agreed in writing first. |
| What does it cost? | Superset is Apache 2.0 licensed with no licence fee. You pay for infrastructure and your engineers' time. | Per Preset's published plans, which include a free tier and a custom Enterprise plan (preset.io/pricing, read 4 November 2026). | A support contract or a quoted managed service. Quote requests are answered within 1 hour. |
| Who answers at 3am? | Your own on-call. The project offers documentation and community help with no response time. | Preset's support for the service, as its plan provides. | An AceMQ engineer within 15 minutes, 24/7. |
| When does it fit? | You have a data platform team that already runs Redis, Celery and PostgreSQL well. | You want a vendor to carry hosting and accept its plans and release schedule. | You self-host for control and want senior help for upgrades, SSO and incidents, or want it run for you. |
For the choice against other BI tools, read Apache Superset alternatives: Metabase, Grafana and Power BI.
One partner, not five
Apache Superset rarely fails in isolation. AceMQ covers the full surrounding infrastructure — so one team owns the whole path instead of pointing at each other.
Guides, articles and comparisons
A guide from AceMQ engineers on when to stay on Superset and when Metabase, Grafana or Power BI fit better, plus the AceMQ services around it: Grafana, Airflow and the databases beneath.
FAQs
Five answers on Apache Superset from AceMQ: who supports it, which versions are covered, what an upgrade involves, when to choose Preset instead and what AceMQ does not do.
Still have questions about Apache Superset? Email an Expert
The Apache project supplies documentation and community help, with no response time. Preset sells a hosted Superset service with its own plans. AceMQ supports the Superset you host yourself, 24/7 with a 15-minute emergency response, and can operate it for you at Tier 3 on your infrastructure or Tier 4 on ours.
Self-hosted Superset 6.x and earlier lines. On 4 November 2026 the latest stable release on PyPI was 6.1.0 (13 May 2026), and 7.0.0 was still a release candidate (7.0.0rc2 on GitHub, 16 October 2026). The project prints no support window per release line, so AceMQ plans upgrades from the release notes and UPDATING.md and tells you what an older version leaves exposed.
Back up the metadata database, read UPDATING.md for breaking changes, run the upgrade on a copy first, then run superset db upgrade and superset init and clear the results cache. The project advises staging first and an off-peak window for production. AceMQ rehearses it and writes the way back down before the window.
Preset is a good fit when you want a vendor to host Superset and you accept its plans and upgrade schedule. Self-hosting fits when data, credentials or sign-in must stay inside your network, or you want to choose versions and add-ons. AceMQ supports the second case and will say when the first suits you better.
No. AceMQ covers the platform: install, upgrades, authentication, performance and operations. Building dashboards, datasets and semantic layers, or rebuilding reports from another BI tool, is content work your analysts or another firm do.
Yes. AceMQ works on Apache Superset in two ways: ad hoc engagements for a defined piece of work, such as a health check, an upgrade or an incident, or an annual contract that covers 24/7 support and proactive reviews. Many teams start ad hoc and move to an annual contract once the first engagement is done.
Every AceMQ proposal is tailored to your requirements: your Apache Superset environment, the SLA you need, the hours of cover and the scope of work. Most contracts run $24,000–72,000 a year, depending on the SLA, the scope and how Apache Superset is deployed, with aggressive discounts when several technologies share one contract. Request a quote and AceMQ replies within 1 hour.
Let's talk
A senior AceMQ engineer reviews your Apache Superset installation, tells you where it stands and what cover fits, and quotes the work. Quote requests are answered within 1 hour.
Production incident right now?Email an Apache Superset Architect
Send us a message and we'll follow up within 1 hour — or book a free 30-min consultation directly.
305-204-2607info@acemq.comPrefer to talk now? Call us directly or use the consultation tab to find a time that works.
