github OneUptime/oneuptime 14.0.16
on GitHub

latest releases: 14.0.17, build-number-9050, build-number-9049...
4 hours ago
Changes
  • Write hooks run only once the caller is checked, and make room only after the write succeeds
  • Every AI agent page puts Test connection and Reset agent in one ⋯ menu next to the agent's status
  • Teams on SSO and SCIM providers are held to what the person saving them could grant
  • fix(references): rules and owner rows accept only their own project's records
  • chore: npm audit fix
  • docs(ai): Ollama setup uses a Base URL the egress guard allows, in all 17 languages
  • An outdated agent version shows a warning sign that opens how to upgrade it, on every resource
  • fix(templates): substitute in one pass so a value's {{...}} text stays literal
  • A runbook's approval steps can't be pre-approved or skipped before the run reaches them
  • A record points only at files of its own project, and pages serve only those
  • fix(monitors): incoming email templates fill the variables the docs promise
  • Create Monitor: what to monitor first, six common types, and default criteria folded
  • Edits take edit permissions: monitoring switch, template settings, workflows and custom domain certificates
  • Every language's runbook page says which step can be completed, approved or skipped
  • fix(templates): expand {{#each}} loops in the same pass as substitution
  • docs(ai): sync the 15 lagging LLM provider guides with English
  • Preview notification is a small 'Preview' link on the line of the value it previews
  • Rename the AI Insights page of every cluster and resource to AI Logs and add a real AI Insights page
  • fix(ai): let the bundled vLLM global provider work with billing enabled
  • feat(ui): a friendlier color picker everywhere - named swatches first, Custom color for an exact code, nothing spilling out of dialogs
  • AI investigations are on by default for every resource, and each Overview ends with its AI agent's status
  • Every service checks that the records it names are the project's own
  • Private images open only to their project; uploads start private and need access to the project
  • AI access follows the agent: green badges, agent-set investigation and fixes, and the commands to change them
  • A paid feature can always be switched off, on any plan: plan-gated settings go back to their default on every plan
  • Templates: Change Monitor Status to and Initial Incident State are editable by the template's editors
  • Hosts, Proxmox, Ceph and VMware agents report their version: outdated sign and upgrade dialog
  • feat(ai): project-wide daily AI token and spend limits under More settings on Project Settings → AI Features
  • fix(status-page): a scheduled maintenance state change with a public note tells subscribers once
  • A reference sent under both of its names must name one record
  • AI daily limit checks read today's AI Logs through an index (online, bounded migration)
  • Incident, alert, maintenance, monitor and status page label and owner rules take the shared two steps; Edit never insists on what a rule adds
  • docs(ai): the LLM provider guide says Global LLM Providers reach private addresses
  • Incidents and Alerts get an AI section with Insights, Logs, Settings and Auto Remediation Rules
  • fix(status-page): a state change's public note is posted or the change is refused, and it names the new state
  • Update side effects follow real changes: re-sent notify flags and severities do nothing, a severity change counts under either name
  • Reference rules hold under either name: service checks, create lists, episode updates, ee/Server guards
  • All/Any only from two conditions everywhere; workspace summaries start weekly and are always scheduled
  • Grouping rules fold everything beyond grouping under More fields (retire Show advanced settings)
  • Who created or changed a record is recorded by OneUptime, read-only in the API and Terraform
  • Create Alert and Create Episode start in the Initial State they offer
  • Subscribers hear about a postmortem once, when it is published
  • Duplicate fills in the copy's name (' 2') and keeps a refused copy in the dialog
  • On-call: each person keeps one palette colour on every card, never black or grey
  • Incoming call escalation rules ring for 20 seconds by default
  • chore: npm audit fix
  • feat(storage-arrays): Storage Arrays product for Pure Storage FlashArray and FlashBlade
  • Incidents, alerts and episodes created already acknowledged or resolved page nobody
  • Say exactly who can turn on SMS, phone calls, WhatsApp and Telegram: a project owner or someone with Manage Billing
  • feat(mobile): ship the iOS critical alerts entitlement in store builds (1.5.0)
  • Heartbeat monitors no longer go Offline when the ingest queue falls behind
  • fix(episodes): clear episode references when members are removed
  • fix(kubernetes-ai): translate the automatic investigation confirmation
  • Serve Farsi docs pages right to left
  • Below a feature's plan, leftover SSO, SCIM, API keys, schedules and Slack/Teams rules can be seen, switched off and deleted
  • Plan and permission checks for Test Rule, a status page's Require SSO and alert/episode state change notes
  • fix(telemetry): coalesced incoming requests evaluate the newest payload
  • Store monitor ingest payloads in jsonb columns Postgres can hold
  • fix(episodes): an incident's or alert's episode follows its membership only
  • fix(api-reference): serve Farsi pages right to left
  • fix(kubernetes-ai): translate the cluster's AI agent page
  • Reads return a record's files only to readers who may see them; status page and form images are public exactly while shown
  • fix(migrations): stop schema migrations queueing ahead of app traffic
  • Workspace summaries go out at the same local time all year: each summary has a time zone
  • fix(permissions): every column is read with its own record's read permission
  • fix(kubernetes-agent): upgrade without --reuse-values, and warn when a release keeps an older chart's defaults
  • feat(kubernetes-agent): opt-in drop of OBI client calls that belong to no trace
  • Label and owner rules: a new rule must add something, everywhere; a rule that only inherits is named after its switches
  • fix(postmortem): a postmortem published while its incident is hidden is sent when the incident is made visible
  • fix(status): readiness fails when a datastore is down; liveness no longer depends on one
  • test(notes): wait for the new event's feed to mount, not just to appear
  • fix(jest): match ignore patterns by whole path segment
  • fix(migrations): keep runtime pods off a schema older than their code
  • fix(ai-access): translate the shared AI access UI into every Dashboard language
  • fix(permissions): every permission check reads team rows by one rule
  • Low-balance messages say who can add balance, and link only for them
  • fix(dashboard): give the Automatic mode card's riskier changes their own translation keys
  • fix(i18n): extract a literal translatableTerm() as a Dashboard translation key
  • Below their plan, API keys and SCIM stop working until the project upgrades
  • Apply incident/alert and episode privacy to episode member rows
  • fix(workflow): clean up schedules after trigger changes and deletion
  • A private incident or episode is never shown on a status page
  • i18n(dashboard): translate the AI agent pages into every Dashboard language
  • feat(navbar): draw the folded product categories as one aligned list
  • feat(dashboard): search and sort the Host Processes list (#4477)
  • feat(workflow): Dynamics 365 / Dataverse workflow templates
  • fix(workflow): Update steps merge custom fields; a reference on its own in a JSON setting is the value
  • Every Send Test asks what sending it for real asks
  • fix(status-pages): what a status page hides stays hidden
  • A plan-gated setting needs its plan on create as well as on update
  • fix(telemetry): every telemetry read follows the caller's label, owned and block scope
  • feat(cloud): discover IaaS and PaaS resources from Azure Monitor, CloudWatch and Cloud Monitoring
  • Billing Admins switch paid channels; Workflow Admins edit workflows, Members run them
  • feat(scheduled-maintenance): Change Monitor Status to can be edited until the event starts
  • Workflow steps act as a Project Admin of their project, on its plan
  • Every diagram runs mermaid built from source, with the katex npm installed
  • fix(feed): alert and episode titles are text in every feed item and chat message
  • fix(github): connecting an installation needs permission to add code repositories
  • Release 14.0.15
Check out latest releases or
releases around OneUptime/oneuptime 14.0.16

Don't miss a new oneuptime release

NewReleases is sending notifications on new releases.

Get notifications