> ## Documentation Index > Fetch the complete documentation index at: https://docs.ainomos.ch/llms.txt > Use this file to discover all available pages before exploring further. # System status and maintenance > See whether everything Ainomos depends on works, plan maintenance, and decide who is emailed and through which mail server. Ainomos depends on several systems, from its database to the models it calls. The **Operations** group in **Settings** helps you watch them and tell people when something is wrong. ## System status Who can see this: platform admins. Open **Settings**, then **System status**. Ainomos checks every system every minute. Mail is checked every 15 minutes, file storage and log shipping every 5 minutes. The top of the page sums up the result, for example "All 12 systems are operational", and links to the next maintenance and to the alert recipients. The systems come in two groups: * **Connected by your organisation**: the models, company sign-in, file storage and log destinations your admins set up. You fix these. * **Shared by the installation**: the parts every organisation on the installation uses. The platform operator fixes these. You see their state. ### What each state means | State | Meaning | | - | - | | **Operational** | The system answers normally. | | **Degraded** | It answers, but too slowly or asking to slow down, on two checks in a row. | | **Down** | No answer or an error on two checks in a row, or real requests failed 3 times in 5 minutes. | | **Maintenance** | A maintenance window covers it. No alert emails are sent. | | **Not checked** | No check has run yet. | A system counts as operational again after two passing checks. ### What people notice when a system fails | System | If it fails | | - | - | | Database | Nothing works until it answers again. Nothing is lost. | | Policy engine | Nothing works. Every permission check is refused, so nobody gets more access. | | Sign-in service | Nobody can sign in. Open sessions keep working until they end. | | Model gateway | Chat, search, imports and dreaming stop. Conversations stay readable. | | Privacy scanner | Rooms that mask personal data refuse questions and imports. Other rooms work. | | Knowledge graph | Rule checks that use the graph, and the Graph page, are unavailable. Chat still answers. | | Mail | Emails through the installation's mail server are not sent, nor kept for later. | | Background jobs | Dreaming and background imports wait. | | Agent connection | AI apps cannot reach Ainomos. People in the app are not affected. | | A chat model | Rooms that use it answer slowly or not at all. | | The search model | Search, imports and dreaming stop. | | Company sign-in | People who sign in through it cannot sign in. Emergency access still works. | | File storage | New imports keep no original, and originals cannot be downloaded. | | Log shipping | Audit events wait and are sent when the destination answers again. | ### Look into one system Click a row to see **What people notice**, **What to do** with a link to the setting that fixes it, the **Last 24 hours** and the **Changes** of the last 30 days. **Check now** checks one system at once, **Check all now** checks all. ### Notices in the app While a system is down, platform admins and maintainers see a red notice for it at the top of every page. Other people never see system names. While the model gateway is down, they see "Answers are not available right now", unless you turn that off under [Alert emails](#alert-emails). In a down system's details, **Silence banner** hides its notice for this outage for everyone in the organisation. It shows again if the system recovers and fails again. **Show banner again** undoes it. ## Maintenance Who can do this: maintainers. A platform admin grants maintainer rights under **Users**. Platform admins who are not maintainers can only read the page. As the page sits in **Settings**, maintainers also need the Platform admin role to open it. Every change on this page is recorded as evidence. ### Close Ainomos now Click **Turn on maintenance mode**, read what it does, and click **Turn on**. * Everyone except the maintainers sees the maintenance screen, signed in or not. * AI apps get a general "server busy, try again later" answer. * Maintainers keep everything and see "You are working as a maintainer". * It stays on until a maintainer clicks **End maintenance**. The maintenance screen then reopens Ainomos by itself. ### Schedule maintenance Click **Schedule maintenance** and fill in the form. **What people will see** previews the notice. | Field | Options | | - | - | | **Level** | **Information**: a blue notice people can close. **Warning**: an amber notice that stays. **Downtime**: a red notice that stays, and Ainomos closes or parts of it pause. | | **Title** and **Message** | The title shows in the notice. The message is optional. | | **Starts** and **Ends** | Starts **Now** or **At** a time. Ends **When I end it** or **At** a time. | | **Show the notice** | For a later start: **From the start**, **1 hour before**, **1 day before**, **3 days before** or **1 week before**. | | **While it runs** | Downtime only. **Close Ainomos**, or **Keep Ainomos open, pause parts of it**: **Chat** (people read but cannot ask), **Agents** (told to try again after the end), **Imports** (new ones wait), **Dreaming** (scheduled runs are skipped). | | **Systems it covers** | They show **Maintenance** instead of **Down** and send no alert emails. | | **Emails** | **Email the alert recipients** when it is scheduled, starts and ends. | Click **Schedule**, or **Start now** if it starts now. The page lists maintenance under **On now**, **Upcoming** and **Past**, with **Edit**, **Cancel** and **End now**. Maintenance the platform operator scheduled for the whole installation is **Read-only**. Two downtime windows cannot overlap. ## Alert emails Who can do this: platform admins. Open **Settings**, then **Alert emails**. By default nobody gets alert emails, and the notice in the app is the only alert. | Recipients | Who it reaches | | - | - | | **All organisation admins** | Everyone with the Platform admin role. The page lists who that is today. | | **Group mailboxes** | Team addresses, so whoever is on duty is reached. | | **People** | Anyone in the organisation, also without an admin role. | | When to email | When | | - | - | | **A system goes down** | Once the next check confirms it, about two minutes after the first failure. | | **It works again** | With how long it was down. | | **A system stays degraded for** | 5, 15, 30 or 60 minutes. The default is 15. | | **Maintenance is scheduled, starts and ends** | For maintenance that asks to email the recipients. | | **Remind while still down, every** | 1, 4 or 24 hours. The default is 4. | **Show people a notice while a system they use is down** controls the notice members see. Admins always see every outage. Outages of shared systems are emailed too. Click **Save**, then **Send a test email** to the saved recipients. ## Mail server Who can do this: platform admins. Ainomos sends alert, maintenance and test emails through one server: 1. **Your own mail server**, when you set one up here. 2. Otherwise the installation's server, shown as **Default mail server**. You can **Send a test email** through it, but not change it. 3. If neither exists, no email is sent, and the page says **No mail server is set**. > [!NOTE] > > Invitations and password emails come from the sign-in service, which has its own mail settings. The operator of your installation sets them. This page does not change them. | Field | What to enter | | - | - | | **Server** and **Port** | The SMTP server, for example smtp.example.ch, and its port. | | **Security** | **STARTTLS**, **TLS** (usually port 465) or **None**. | | **Username** and **Password** | Optional. The password is stored encrypted and never shown again. | | **From address** and **From name** | The sender. The server must allow sending as this address. | | **Test email to** | Where the test email goes. | Click **Test**. It checks **Reach the server**, **Secure the connection**, **Sign in** and **Send a test email**. **Save** becomes available after a test that passed. In the Ainomos cloud, servers at private network addresses are refused. **Remove** goes back to the installation's server, or stops alert and maintenance emails if there is none. Saving and removing are recorded as evidence. ## Related * [Settings](https://docs.ainomos.ch/admin/settings-overview.md) * [Users](https://docs.ainomos.ch/admin/users.md) * [File storage](https://docs.ainomos.ch/admin/file-storage.md) * [Log shipping](https://docs.ainomos.ch/privacy/log-shipping.md) This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.