Skip to content

Incidents

Incidents & Maintenance 17 min read
On this page

An incident is Uptimia's record of one problem on one monitor: when it was confirmed, what the error was, which probes saw it, how long it lasted, and who was alerted. Most incidents open and close automatically. When something is wrong that the checks cannot see, you can also open a manual incident yourself (see Manual Incidents below). Individual check results, one row per check per probe, live in Monitoring Logs.

How Incidents Are Created #

An automatic incident opens when a monitor confirms a problem and closes the moment it passes again. All twelve families produce incidents, each with its own word for what one is. "Down" appears only where a service stopped answering.

Monitor familyWhat opens an incidentState word on monitor pages
UptimeHTTP error, timeout, connection failure, keyword mismatch, port or protocol failureDown
TransactionA step in the recorded journey failedDown
APIA request or assertion in the chain failedDown
HeartbeatNo ping arrived before the schedule plus grace expiredDown
ServerA CPU, memory, swap, disk, inode, load, process or network rule breached, or the agent stopped reportingDown
SpeedPage too slow, Vitals over budget, a tracked element failing, or the page unreachableSlow · Vitals over budget · Element failing · Down
SSLCertificate expired, chain or protocol problem, or the host unreachableCertificate problem · Down
DomainRegistration expiring or expiredExpiry warning
DNSA record changed against the baseline, or nameservers failed to answerAlert active
SecurityA finding at or above the monitor's Alert on level: malware flagged by Google Web Risk, or an issue in another check group such as software versions, security headers or TLSWarning, with the finding on the line below; a malware verdict reads Infected
BlacklistThe domain or IP address appeared on a spam blocklistListed
Real User (RUM)Traffic stops, traffic volume drops, average speed drops, JavaScript-error rate spikesDown

Speed and SSL each produce more than one kind of incident, told apart by the error they report: a certificate problem and a host that never answered are separate events, and the wording follows the error. The incidents pages label the Real User family RUM.

Confirmation Before an Incident Opens #

For Uptime, SSL, Speed, Transaction and Domain monitors, one failing check is not enough. Uptimia dispatches backup checks from other, distinct probes. The incident opens only once enough of them also report down. The Outage confirmation setting is per monitor, on the editor's Advanced tab, range 1–3, default 3, and recovery runs faster than detection: one passing backup check recovers the monitor whatever the threshold is. Security, Blacklist, Heartbeat, Server and Real User monitors have no such control. API monitors count the primary run as one of the observations, so a threshold of 1 opens the incident from the first failing run. DNS skips the confirming round at 1. The full per-family counting rules are in How Uptimia Monitoring Works.

Severity, Maintenance and Paused Monitors #

Every incident carries a severity: critical or degraded. The incident lists, the status badges and the incident page's own severity badge all use these two words. Only critical incidents count as downtime in your availability figure; a degraded incident stays in the lists but adds no downtime. An automatic Security incident is critical and alerts, but it is a finding, not an outage, so it never counts as downtime either. To stop a critical incident counting, or to count only part of it, set its availability impact (see Availability Impact below). The vocabulary is in Monitor Statuses.

A check result that arrives while a monitor is inside a maintenance window is dropped on arrival: no log row, no incident, no alert. See Maintenance Windows.

Paused monitors are excluded from both account-wide incident pages. An open incident on a paused monitor stays open and stored, hidden here until you unpause. The monitor's own Incidents card still lists it.

Active Incidents #

Go to Incidents → Active. The page lists everything down or degraded right now, across every monitor type, and it refreshes every 30 seconds, as the line under the heading says. Each row's duration ticks up every second. Four tiles sit above the list:

TileWhat it shows
Active incidentsTotal open incidents, with a N critical · N degraded split underneath
Monitors affectedDistinct monitors failing, N of M against your fleet, broken down by type
Longest ongoingThe live duration of the oldest open incident, and its monitor name
UnacknowledgedOpen alert groups nobody has acknowledged, N of M
The Active Incidents page with its four KPI tiles, the type and monitor filters, a correlation banner, two alert groups (one awaiting acknowledgement, one acknowledged) and the live list of every other monitor that is down or slow, each row ticking its own duration.Enlarge

Filtering the Active List #

Filters are one row above the list.

  • Search matches the monitor name or the error text.
  • Any type: a dropdown listing All plus each type that currently has an incident, with its count.
  • Specific monitors: a pill dropdown grouped by type, a checkbox per monitor, All monitors at the top to reset.
  • Sort: Newest first, Oldest first or Longest down.

All four apply instantly, with no page reload.

The Correlation Banner #

Sometimes three or more open incidents, alert-group members included, report the exact same error inside a 3-minute span, and an amber banner then appears above the list: "N monitors probably share one cause". It names the monitors and the span, and Highlight marks the member rows. One shared cause is the usual explanation: a resolver, an upstream provider, a shared load balancer.

The amber correlation banner over the active incident list, telling the reader that three incidents share one error and one short window, with the matching rows highlighted below it.Enlarge

Alert Group Cards #

Incidents on monitors that have an escalation policy attached are wrapped in an alert group and shown as one card, headed "N of M monitors down". The card carries the policy name, a line such as "3 notification rounds instead of 10", an Acknowledge button and a View incident link to the group page. The whole storm gets one digest, one ladder and one recovery alert.

Creating the escalation policy that produces alert groups needs the Professional plan or above, though a trial clears that gate for its whole run, and a policy created earlier keeps grouping and acknowledging after a downgrade. On an account with no escalation policy, and on any monitor set to All at once, every incident gets its own row. The mechanism is Alert Groups; the workflow is Acknowledging Incidents and MTTA.

Rows in the Active Table #

The list is split into sections. Needs acknowledgement and Acknowledged hold the alert groups, Down holds critical incidents on monitors with no escalation policy, Warning holds Security incidents on those monitors, and Slow, not down holds degraded ones. A group held open by a maintenance window sits in In maintenance.

Each row carries a severity dot (red critical, amber degraded or Security warning), the monitor name and type, the error, the time it started (shown as since <timestamp>), the live duration and how many alerts were sent. Clicking the row opens the incident. The kebab menu holds View details, Post status update and Open monitor. There is no Delete: open incidents cannot be deleted.

Incident History #

Go to Incidents → History. History shows every incident that overlaps the selected period, resolved and ongoing alike: an incident that started before the window and ran into it is included, and its bar has an open left edge. Its four tiles differ from the active ones:

TileWhat it shows
IncidentsTotal matching the window, counted across every page
Total downtimeSummed critical incident time, and how many monitors it spans
Mean time to recoveryAverage duration of the resolved incidents, and how many were averaged
Still ongoingIncidents with no end time yet, and when the oldest started

Total downtime and Mean time to recovery cover only the incidents on the current page, and say so when more pages exist. Widen the range and the page size before reading them as account totals. Still ongoing counts every ongoing incident in the range, because they are all listed on page 1. Total downtime leaves out any time an availability impact setting excludes.

Your plan sets the monitoring-history window: Free 7 days, Trial 30 days, Basic 90 days, Professional 365 days, Enterprise 730 days. Nothing is deleted at that boundary. Upgrading widens what you can read immediately and retroactively. The incident pages do not clamp the range you pick, so a wider range can reach further back here than the charts on a monitor's own page. See Data and Log Retention.

Filtering History #

The date range picker comes first in the filter row: the window drives everything below it. Next come a search box, the status chips All · Ongoing · Resolved with counts, an All types dropdown listing only the types present, and the Specific monitors pill. Every filter runs on the whole result, not only the page you are looking at, and any change takes you back to page 1. The status chips and the type dropdown show counts for the whole result.

The Downtime Timeline #

Above the list, a per-monitor swimlane draws the range as horizontal bars: red critical, amber degraded, one lane per monitor, sorted by total downtime, with the top seven lanes showing and a + N more monitors expander underneath. Hovering a segment gives the error and times. Clicking it opens that incident. Day gridlines and a legend (Operational · Down · Degraded) run below. When history spans more than one page, a line under the chart states which incidents it reflects.

The Incident History page with the downtime-by-monitor timeline above the day-grouped incident list, a checkbox on each row that can change availability impact, the Duration column, and the still-ongoing incidents pinned at the top.Enlarge

Reading the History List #

The list is grouped by the day the incident started, newest first, with a Still ongoing group pinned above everything on page 1, however many ongoing incidents there are. Later pages hold resolved incidents only. Each day header carries its incident count and downtime subtotal.

ColumnWhat it shows
Severity dotRed critical, amber degraded or Security warning, gray once resolved
MonitorName (links straight to the monitor) with its type underneath
ErrorThe error that opened the incident, as a badge
Started → endedstart → end, or start → now while it is still running
DurationThe final duration, or a live-ticking counter while ongoing
AlertsThe delivery tally, such as 3 accepted · 1 failed (parts that are zero are left out), or N alerts sent for an incident with no tracked deliveries

Rows from an alert group carry a note under the monitor name: who acknowledged it and how fast ("Acknowledged in N min by X"), or "Group #id · not acknowledged, recovered before step N". An incident from a multi-monitor group also shows "N notification rounds instead of M" in the alerts cell. The per-recipient breakdown lives on the incident page's Timeline card.

The kebab menu holds View details, Post status update, Open monitor and, on ended incidents, Delete. History paginates at 50 rows per page, switchable to 20 or 100.

Owners, Admins and Editors also get a checkbox on each row that counts as downtime. Ticking rows opens a bar with Don't count (asks for a reason) and Count in full, which change the availability impact of every selected incident at once. A row whose impact is changed carries a badge such as Doesn't count or Counts 12 of 65 min; hover it for the reason.

The Incident Page #

You can open an incident from a row, from the kebab's View details, or from a swimlane segment. Each one opens as a full page of its own rather than a slide-out drawer. An incident on its own opens that incident's page; an incident inside an alert group opens the group's page, where the whole storm is described together.

The page opens with the monitor name, the incident number, and a status pill reading Ongoing or Resolved. While it is ongoing, a Speed incident's pill reads Critical or Slow, and an SSL or Domain incident's reads Critical or Warning. If the monitor carries notes, they appear in a Your note card beside the timeline, so whoever is responding sees them without leaving the page. A summary card follows: the error with its severity and how many locations failed, then Started, Ended, Down for and Alerts.

Below that, the page shows only the evidence cards the incident has data for:

CardWhat it holds
Where it failedOne row per location: flag, IP address, when it checked, the result and the response. Locations that ran no check sit behind a line under the table. While the incident is open the card is titled Where it is failing.
Transaction stepsPer-step results with the failing step marked, plus step screenshots
What the monitor sawEach distinct error with an occurrence count, the screenshot at failure where the family produces one, the response headers, response body and traceroute (each collapsible), and the keywords the check was matching

Alongside it, a Timeline card lists which contacts received which alerts and when: the incident opening, each alert with its channel and destination, and the recovery. Security, Real User, server, heartbeat and blacklist incidents record no detecting probes, so they show no Where it failed card. SSL, DNS and blacklist incidents get a card of their own for what the check found. An automatic Security incident shows a What we found card: the finding that opened the incident, marked Opened this incident, then any other finding open during it, each with its severity, evidence, how to fix it and whether it is still open, cleared or marked as expected. A speed incident shows Load time by location instead, plus a Locations without a load time card when some locations failed to load the page. No incident page carries a response-time chart.

A resolved incident page showing the status badge, the error and timing strip, the Where it failed and What the monitor saw cards, and the timeline that lists which contact received which alert.Enlarge

Actions on an Incident #

The actions sit in the page header, next to the incident name. Availability impact and Delete incident are in the ⋯ menu.

ActionWhat it doesWho sees it
Export → Download PDFThe Root Cause Analysis as Root-Cause-Analysis.pdfEveryone
Export → Download HTMLThe same report as Root-Cause-Analysis.htmlEveryone
Export → Download alert deliveries (CSV)One row per alert delivery for this incident: channel, recipient, status, attempts and errorsEveryone
Export → Create public linkA URL that opens the report without signing in. Not offered for an incident set to not count at allOwner, Admin, Editor
Post status updateA status-page update pre-filled from this incidentOwner, Admin, Editor; Uptime, Speed and Transaction only
Open monitorJumps to the monitor's detail pageEveryone
Resolve manual incidentCloses an open manual incident, after a confirmationOwner, Admin, Editor; open manual incidents only
Availability impactSets how much of the incident counts as downtimeOwner, Admin, Editor; incidents that count as downtime
Delete incidentRemoves the record permanently, behind a confirmationOwner, Admin, Editor; only once the incident has ended

While an incident is running, the ⋯ menu says Delete becomes available once the incident ends in place of the button. On monitor types other than uptime, speed and transaction, the Post status update button is disabled and explains why: "Status updates are available for uptime, speed and transaction incidents."

For Owners, Admins and Editors, every working public link for the incident is listed under the page header with Copy and Revoke. A revoked link shows a page-not-found error to anyone who opens it, and you can create a new one afterwards.

Warning: Deleting an incident cannot be undone, and it removes the incident from your history, your exports and your availability record. Any status page thread linked to it is deleted too, and its public links stop working.

Availability Impact #

Some incidents should not count against your availability: a failover you planned, or a false alarm caused by your own firewall. Open the incident, choose ⋯ → Availability impact, and pick Counts in full, Doesn't count or Count part of it (with Count from and Count until times inside the incident). A reason is required. The incident, its alerts and its logs stay as recorded; only the downtime figure changes, on the dashboard, monitor pages, reports and public status pages. While a change is set, the page shows an Availability impact card with the outcome, the reason and a Change button. Changes are logged in the audit log and can be undone.

Setting Doesn't count or Count part of it needs the Professional plan or above (trials included). After a downgrade, existing settings keep applying and you can still restore an incident to count in full. Incidents that never count as downtime, such as degraded ones and automatic Security incidents, do not offer the option.

The Availability impact dialog on a resolved uptime incident, with Count part of it selected, the Count from and Count until times, a reason typed in, and the counted downtime of 4m 5s out of 6m 59s.Enlarge

Exporting Incidents #

The whole filtered list. On History, click Export CSV. The file is UptimiaIncidents.csv. It honors the date range, the monitor filter, the search, the type and the status and, like the page, excludes paused monitors. Columns at the end record who opened a manual incident, whether a manual incident closed the incident, and each incident's availability impact, counted seconds and impact reason. The Active page has no CSV button, so export from History. The columns are listed in Exporting Your Data.

One incident. Use the Export dropdown on the incident page for PDF or HTML, or Create public link for a link that anyone can open without an Uptimia account. Turning these into a write-up is Running an Incident Postmortem.

Manual Incidents #

A manual incident records a problem that you report rather than the checks: checkout payments failing, an API that is slow for some customers. Choose Open manual incident from the monitor's actions menu on the dashboard or on its detail page. Every monitor type except Real User supports it, and Owners, Admins and Editors can open one.

  • Write a message of up to 500 characters. It appears in the alert and on the incident page.
  • Choose Critical (the monitor shows Down and the time counts as downtime) or Degraded (the monitor shows Degraded, no downtime).
  • On uptime, speed and transaction monitors that appear on a status page, tick Also post to status pages to publish the message as a public incident.

The alert goes out at once, ignoring the monitor's alert delay; anyone outside their alerting hours gets it when those hours start. While the manual incident is open, automatic judging for that monitor stops: checks and logs keep running, but no automatic incident opens or closes. An automatic incident already open is closed with no recovery alert and replaced by yours. A banner on the monitor page and the dashboard starts Automatic alerts paused by a manual incident, and lists carry a Manual tag. Only a person ends it, with Resolve manual incident on the incident page or Resolve on the banner. If it stays open for 24 hours, the person who opened it (or the account owner) gets a reminder email, repeated daily.

A paused, suspended or in-maintenance monitor cannot get a manual incident. A maintenance window that starts later does not close one.

The Open manual incident dialog for an uptime monitor, with the message box and its character counter, the Critical and Degraded severity choices, and the Also post to status pages checkbox.Enlarge

Going From a Monitor to Its Own Incidents #

Most monitor detail pages carry an Incidents card listing the eight most recent incidents. Its footer link, View all incidents, opens Incidents → History with the monitor filter already applied. On the Server page the card lists ten instead of eight and labels its link View all N incidents, while the Real User page lists its incidents in a full table with no link.

Below the Incidents card, a log card does the same for check results, opening the logs page pre-filtered to that monitor, headed Check log on most families, Sweep log on DNS and Ping log on heartbeat. Every family's page carries both cards except Real User. Unlike the account-wide pages, the Incidents card keeps listing incidents while the monitor is paused.

Was this helpful?

Your answer helps the team improve this article.

Still need help?

Uptimia Bot answers right away and passes you to the team when it can't.

Source: https://help.uptimia.com/help/a/incidents-monitoring-logs