Varqeni Help

What's new

Updated 11 November 2026

A short round-up each month of what changed for people using Varqeni, with a link to each guide.

November 2026

Incidents on your Lock Screen

In Varqeni for iPhone 1.1.2, an incident that pages you stays on your Lock Screen and in the Dynamic Island until it is resolved, with Acknowledge and Open. A widget shows your open incidents and next shift, and Approve asks for Face ID.

Incidents on your Lock Screen, and the Varqeni widget

A phone call when nobody answers a page

Mark an escalation step to call, and a page that reaches it phones the people on that step with an AI voice that briefs the incident. Say acknowledge to take it, or pass it on. Live in the Fortis lab.

Voice calls when nobody answers a page

Varqeni for iPhone and iPad

The Varqeni app is in the App Store. Connect it to your console and it notifies you when you are paged, shows what needs you, and lets you acknowledge, escalate, resolve, or approve or decline a fix, as your role allows.

The Varqeni app for iPhone and iPad

October 2026

A clearer investigation page

Each investigation now opens like a problem page: its state and numbers at the top, the cause with its confidence and numbered sources you can open, every cause Varqeni considered, where it sits on the live map, the steps on one time axis, and what fixed the same thing before. Tell Varqeni whether it was right with one click.

Causes, citations and steps

Home numbers, the Investigations list and shortcuts

Four numbers on the home page, every investigation in one filterable table that says what Varqeni already did, keyboard shortcuts, and a health check for your connections.

The Investigations list

What Varqeni may do, in plain words

Policies lists every action as runs on its own, asks first or analyze only, with a ceiling per system and one switch to stop every action. Plus a weekly brief for leaders and an MCP server for coding assistants.

What Varqeni may do on its own

Single sign-on, directory sync and an audit log

Varqeni can sign people in with Entra ID, Okta, Google, GitLab or any OpenID Connect provider, keep people and teams in step with your directory through SCIM, and show who did what in one exportable audit log. Built and tested; single sign-on and SCIM are not yet tried against a real directory.

Single sign-on

The world model

Everything Varqeni watches in one picture: every part and every machine under it, grouped by type and coloured by health, with counts of what needs a look.

World model

Status page from the console

Set up your status page in the console: its title and components, with 90 days of uptime and an Atom feed of updates.

Set up the status page

Users and access in the console

Admins now add people, give them a role everywhere or on chosen systems, group them in teams and make service accounts, all from the console, with every change kept in Settings history. The gear opens Administration; data sources stay under Connections. Live in the Fortis lab.

Users and access

Error budgets

Give a service an objective and Varqeni pages when its error budget is being spent fast, and shows a slow burn without paging. 30 of 30 blind lab trials.

Error budgets for your services

Steadier investigations and pages

A page Varqeni opened now closes when the system recovers even if Varqeni restarted in between. When a service logs errors only because a service it calls is failing, Varqeni names the service it calls, not the one echoing it. The Summary, Timeline, Evidence and Audit trail tabs of an investigation stay in reach at every window size, and the help centre shows the console as it looks today. Live in the Fortis lab.

Read an investigation

Install with one command

A single varqeni file installs Varqeni as a service on Linux, with the console inside it, then connects one part of your estate at a time and switches each on only after a read-only check passes. Built and tested on fresh hosts, not offered yet.

Install with one command, then connect what you run

Dashboards and Explore, built in

Every system Varqeni watches now has dashboards that build themselves from your Prometheus, with incidents, deploys and fixes marked on every chart. Duplicate one to make it your own: every save is a version you can go back to, and two people saving at once never overwrite each other. Explore queries any metric and turns it into a panel or an alert rule. Tested live in the Fortis lab.

Dashboards

Resources

Every machine in your estate on one page, laid out the way Proxmox lists it: what each is given and uses, both memory figures, what runs on it and what Varqeni thinks of it, with a page of its own for each that a link opens directly. Live in the Fortis lab.

Resources

Settings

Every data source and integration in one place, with a Test button. Add a connection from a catalogue with its setup guide; Save & test checks it before anything is kept. Every change is a version you can compare and roll back, and secrets are never shown again once saved. Tested live in the Fortis lab.

Settings: connections

Data volume and schema drift

Varqeni now tells you when change data stops flowing at its usual volume, or the source schema changes, even while the pipeline looks healthy, and names a schema change with the fault that follows it. 30 of 30 blind lab trials.

Data volume and schema drift

A console made for the job

The Overview opens on a greeting and your pipeline drawn live: each part with its product mark, its one key number and its recent history, the rate moving between parts, and the platform it runs on as one strip of lights. The estate is an orbit around Varqeni, the live map goes full screen, and the console is now dark only, like most tools people watch all day.

Read the live pipeline

Look back with the time bar

Drag the needle on the time bar to any moment and every system redraws as it was then, with the day or week before to compare and a link that opens on the same moment. Reports now show trends over weeks, each with a CSV download. Live in the Fortis lab.

Look back with the time bar

OpenTelemetry logs and metrics

Service Operations now names a service that still answers but logs errors, and one whose own metric crosses a rule you set, from your OpenTelemetry Collector. It reads only a count of error logs, never their text. 35 of 35 blind lab trials.

OpenTelemetry logs and metrics

Schedules, overrides and escalation policies

On-call can now run several schedules, let people cover for each other from the console, and send each page along an escalation policy chosen by system, cause and source, repeating until someone answers. An override and routing were tested live in the Fortis lab; repeat escalation and handoff notices are tested in code.

Schedules, overrides and escalation policies

Running an incident

Declare an incident Varqeni has not seen, set its severity, take command, post status updates to the incident's thread and keep action items, all from the console and checked against the on-call rota. Tested live in the Fortis lab.

Run an incident: declare, severity, command, updates and action items

Reports

Time to answer a page and to resolve an incident, the noisiest causes and each person's on-call load, across every system.

The Reports view

Alert rules that come with a diagnosis

Write a rule on any metric, guided or in PromQL, checked against your Prometheus as you type and replayed over the last week before you save it. When it fires, Varqeni investigates and pages with the cause and evidence, not just the number. Tested live in the Fortis lab.

Alert rules that come with a diagnosis

Service desk

Requests and problem reports in one queue, with an ITIL priority from impact and urgency, response and resolution clocks, an owner from the on-call rota, and a link to the incident Varqeni already has open: when it recovers, the ticket resolves with it. Tested live in the Fortis lab.

Service desk

Deploy events and a read-only API

Your CI can tell Varqeni what it deployed: a service that starts failing soon after is named as the bad deploy, with the version, commit and who. A read-only API with tokens lets your own tools read incidents and their timelines. Both tested live in the Fortis lab, with a signed outbound webhook that sends every notice to your own endpoint and retries when it is busy.

Webhooks, deploy events and the read-only API

Alerts from your monitoring

Alerts in Prometheus Alertmanager's webhook format now open incidents in Varqeni, are investigated like a fault it found itself, and resolve with the alert. Tested live in the Fortis lab.

Alerts from Prometheus Alertmanager and other monitoring

The seal checks the locked copy

The seal on an investigation now compares the copy you read with the one locked in object storage, and Varqeni will not act on an incident whose copy differs. Who acknowledged or approved a page is chained and locked the same way, and Varqeni backs up its own state every night. Tested live in the Fortis lab.

The audit trail and the verified seal

Silences and maintenance windows

Hold pages back for a system while you work on it, now or as a window later, from the On-call view. Every incident it covers says so in its timeline, and the page goes out if the fault is still there when it ends. Tested live in the Fortis lab.

Silences and maintenance windows

Pulse, and tracing on the live map

The Overview now gives every system one pulse row with its key signals over the last three hours, read from your Prometheus; point at a line to read any moment. On the live map, choosing a part traces everything upstream and downstream of it, and a fault is brought into view and traced on its own.

The live map

A new console: the live map and incident stories

The console opens on one sentence about what needs a person across everything Varqeni watches, with the fixes waiting for you, live signals and what Varqeni did on its own. The live map shows each pipeline, your services, the platform under them and the hosts in one picture. Any incident can be played as a story, step by step.

The live map

A limit on actions an hour

Each system now takes at most six actions an hour, whoever approves them: people, the autopilot and runbooks together. One more is refused before anything is sent, and the person on call is paged. Tested live in the Fortis lab.

Acknowledge, approve, decline or escalate from the console

The data check runs by itself

After every pipeline incident recovers, Varqeni now checks on its own whether any rows were lost or duplicated, and tells the team only when they were. Tested live in the Fortis lab.

Check an incident for lost or duplicated data

Evidence kept for one year

Every investigation is now kept write-once for one year, up from 30 days.

How evidence is sealed and kept

PostgreSQL health

Varqeni now watches the PostgreSQL server itself, not only the slot a pipeline reads: long and idle transactions, replication slots nobody reads, connections near the limit. It recommends and does not act. 30 of 30 blind lab trials.

Read PostgreSQL health

The platform underneath a data incident

A pipeline investigation now names what broke underneath it, such as a Kafka Connect worker with no node to run on, beside the cause. 32 of 32 blind lab trials.

The platform underneath a data incident

Runbooks with several steps, shared across pipelines

A runbook can now wait, check the system is still failing, run its fix and check it stays healthy, paging a person if any step fails. One runbook can be shared by several systems of the same kind. Tested live in the Fortis lab.

Custom operations: write a runbook

Debezium Server

Varqeni now watches Debezium Server, Debezium without Kafka Connect, as a second change data capture engine: the engine's health, its slot, heartbeat and topic. It recommends and does not act on it. 30 of 30 blind lab trials.

Debezium Server

OpenTelemetry traces

Service Operations now reads request errors per service, and which service calls which, from an OpenTelemetry Collector. A dependency seen in the traces counts like one written down, so a failure is named once, at its source. 30 of 30 blind lab trials.

OpenTelemetry traces

Kubernetes autoscaling and load balancers

Infrastructure Operations now names an autoscaler held at its maximum while it still wants more, one that cannot read the metric it scales on, and a load balancer with no address or no ready backends. A crash loop behind a load balancer stays the cause. Read-only. 30 of 30 blind lab trials.

Infrastructure through Prometheus

Pipeline watch follows deploys and task restarts

A new release of an application that writes to your source now starts a watch, like a connector or schema change. The baseline also counts how often the connector's tasks restart, and a burst of restarts during a watch is drift. Tested live in the Fortis lab.

Pipeline watch

Paging through PagerDuty

Varqeni now pages through PagerDuty as well as Teams: one PagerDuty incident per fault, however long the fault lasts, resolved when the system recovers. Live against the Fortis lab's own PagerDuty account.

PagerDuty

Kafka broker health and topic freshness

Infrastructure Operations now reads Kafka's own view of itself: a broker down or fenced out of the cluster, no controller, offline or under-replicated partitions, and a topic that stops receiving messages. A broker outage is named once, with the partitions it leaves short as consequences. Read-only. 30 of 30 blind lab trials.

Infrastructure through Prometheus

Acknowledge and approve from the page in Teams

With the Varqeni app in your team, a page carries Acknowledge, Approve with a confirm step, Page the next person and Decline. Teams tells Varqeni who pressed the button, the on-call rota decides whether they may, and the outcome follows in the thread. Tested live in the Fortis lab: a fix approved from Teams ran and was verified in about 5 seconds.

Notices in Microsoft Teams

Check an incident for lost or duplicated data

After a pipeline incident, the person on call can ask Varqeni whether any rows went missing or arrived twice. It compares the primary keys in the source with what reached the topic, read-only, and names the missing ids or says a second snapshot sent the duplicates. Right in 24 of 24 lab trials against real faults.

Check an incident for lost or duplicated data

Request error rates and dependencies between services

Service Operations now reads the share of real requests each service fails, so a release that breaks only some requests is caught even when a single check passes, and knows which services call which: a failure passed on by a caller is named once, at the service where it started, so a rollback is aimed at the release that broke. 42 of 42 blind trials, fixes 32 of 32.

Read your services under Service

Service Operations is live in the lab

Varqeni now watches the services your users reach, probed from outside, with their scheduled jobs and recent deploys. It names the cause when one breaks (30 of 30 blind trials) and, after a named approval, rolls back a bad deploy or restarts a hung workload (16 of 16 closed-loop trials, the wrong fix never sent).

Read your services under Service

Infrastructure fixes, after approval

Infrastructure Operations can now fix what it finds, after a named approval and only on hosts and namespaces your administrator lists: restart a host's metrics exporter, clear old rotated logs from a full disk, restart a hung workload and roll back a bad rollout. They passed 32 of 32 closed-loop fault trials, and the wrong fix was never sent.

The fixes Varqeni can run

Pipeline watch and Custom operations

After a connector configuration or schema change, Varqeni now watches the pipeline for an hour against its own baseline and says if it drifts; ask PelQi to watch it for longer. Runbooks turn a cause into one known fix, approved each time or, after their own trials, run on their own.

Pipeline watch

Incident timelines and write-ups

Every investigation now has a Timeline, from the first signal to recovery with who did what, and a Write-up: a blameless post-incident review drafted from the record, which you can copy, download, or file as a follow-up in Jira.

Post-incident write-ups and follow-ups

On-call paging, and approvals in the console

Varqeni now pages the person on call only when it cannot settle an incident itself: a fault the autopilot owns is held for its fix first. Pages mention the person on call in Teams, escalate through your rota when nobody acknowledges, and resolve on recovery. Acknowledge, approve or decline the recommended fix from the incident in the console.

Paging and escalation

Help centre and help in the console

Guides for every part of Varqeni, with search, are here at varqeni.com/help. The console links to the article for the screen or error in front of you, and can now be installed as an app from Chrome, Edge or Android.

Get help from inside the console

Sign in from anywhere, with a security check

The console is reachable from the internet at varqeni.fortisdatalab.com, behind the same sign-in. Sign-in now has a Cloudflare security check, and attempts from the internet are counted apart from Fortis's own network.

Sign in to the console

Know when a downstream reader stops

Varqeni now spots a service reading your change data from Kafka that has stopped or keeps falling behind, tells your Teams channel once, and shows it on the pipeline. It reports this; it never restarts readers.

A downstream reader stopped or is falling behind

The autopilot fixes two known faults within your policy

Under a written policy, Varqeni now resumes a paused connector and restarts a task left failed after an outage without waiting for a person, checks the fix worked, and tells Teams. Each passed its own closed-loop trials first.

The autopilot: acting within policy

Infrastructure Operations in early access

Varqeni now reads hosts, virtual machines, Kubernetes, storage, data services and certificates from the metrics your Prometheus already collects, names the cause of related problems, and recommends. It changes nothing.

Read the estate under Infrastructure

Actions, Policies and Systems views

Three new views in the console: every fix and who approved it, the rules Varqeni works within, and every pipeline it watches. A Data and Infrastructure switch sits beside them.

The Actions view

Ask PelQi on phones, and Copy

Every PelQi answer has a Copy button, and on phones the question box stays above the keyboard.

Ask PelQi a question

Still stuck? Contact support or email info@fortisdatagroup.com.