Frequently asked questions

Yes. 30 monitors at 5 min intervals, one status page and 3 months of history, with no time limit and no card required. It covers HTTP(S), keyword, ping and port checks and alerts by email, Discord and Google Chat. JSON/API, DNS, UDP, cron heartbeats, SSL and domain expiry, change detection, faster checks and Telegram alerts start on Starter.

A failure must be confirmed from all two regions before we open an incident and alert you. Live regions today: Amsterdam and London. One region with a bad route cannot page you.

Servers are hosted in the EU.

Yes. Paste a read-only UptimeRobot API key or upload the CSV export. HTTP, keyword, ping, port and heartbeat monitors are recreated with their intervals (raised to your plan minimum where needed) and paused ones stay paused. On the Free plan, monitor types that start on Starter (such as heartbeat) are skipped and marked in the preview. Your alert contacts are listed for you to re-add, and you see a preview before anything is created.

No, by design. Alerts go to email, Telegram, Slack, Microsoft Teams, Discord, Google Chat, Mattermost and signed webhooks. If a call must wake someone, send the alert on to PagerDuty (Agency plan). Telegram uses the shared Tenium bot and starts on the Starter plan.

There is no trial period. The Free plan never expires, so you can test the basics for as long as you like, and the first payment on any paid plan has a 14-day refund.

Yes. A plain JSON API with bearer keys: read keys on every plan, write keys on Pro and Agency. Heartbeat URLs and signed webhooks work without a key. There is no SDK or OpenAPI file yet, and every example is plain curl.

On published list prices, Pro at $29 is about 35% cheaper than UptimeRobot Team (about $44). Every comparison page shows the arithmetic. We also sell agency features most tools do not: client workspaces with client logins and per-client branded reports.

No. Seats are included: 1 on Free, 2 on Starter, 5 on Pro and unlimited on Agency. Client viewers on Agency are free and unlimited.

One URL, host, port, DNS record or domain being checked. An SSL check on an https monitor is included in that monitor. Heartbeats have their own allowance and start on Starter.

Yes, 14 days no questions asked on any first payment.

Yes. An upgrade applies immediately and you pay the new price minus the unused value of the current period. A downgrade or a cancellation takes effect at the end of the period you paid for. Switching from yearly to monthly happens after the yearly period ends.

No VAT is added to the prices. Each payment gets a numbered invoice, listed on the Billing page. Your company name and VAT ID are optional; if you add them they are only printed on the invoice.

Your plan stays active for a grace period of up to 7 days and we email you a payment link. After that the account moves to Free; monitors above the Free limit are paused and come back when you pay.

You cannot create more monitors until you upgrade or ask us for a custom plan. Existing monitors keep running.

There is no trial period. The Free plan never expires, so you can test the basics for as long as you like, and the first payment on any paid plan has a 14-day refund: if it does not fit, you get the money back.

Starter or above. Telegram uses the shared Tenium bot, so you do not create a bot of your own. Free includes email, Discord and Google Chat. SMS and voice-call alerts are not offered; PagerDuty (Agency) can phone someone.

Servers are hosted in the EU. A data processing agreement is available on request; the current draft is on the DPA page.

Yes. Import your monitors in one click with a read-only API key, or upload the CSV export. Check the preview and start; your alert contacts are listed for you to re-add. Nothing at the source is deleted.

After the referred customer makes their first successful payment. Signing up, starting a trial or staying on Free does not trigger anything. Both credits are applied together, and both accounts see the entry in their billing history.

No. It is account credit that is deducted from future Tenium invoices. It cannot be withdrawn, transferred to another account or exchanged for money.

No. Referring your own other account does not earn credit. This programme is for real introductions between different people.

There is no cap on the number of referrals in these rules. The cap applies per referral: credit for one referred customer never exceeds $29, for the referrer and for the new customer alike.

Referral credit is for customers who tell a colleague. The affiliate programme is for people with an audience, such as agencies with a blog, educators or hosting companies, and it pays a recurring percentage in cash. The two are separate and you apply to the affiliate programme on its own page.

Email, Discord, Google Chat, Slack, Telegram, Microsoft Teams, Mattermost, signed webhooks, Zapier and PagerDuty. That is the full list: there is no SMS, no voice call and no mobile app.

No. Slack, Teams, Mattermost, Discord and Google Chat use a webhook URL you paste. Tenium can post to that channel only and cannot read your workspace.

Email, Discord and Google Chat are on every plan including Free. Slack, Telegram, Teams and Mattermost start at Starter ($7/month). Webhooks start at Pro ($29). Zapier and PagerDuty are on Agency ($59).

Press Send test next to the contact in Settings, Contacts. A message labelled TEST is delivered within seconds. You need at least one monitor, because the checker dispatches alerts. The demo on this page is a simulation and sends nothing.

Yes. Add one contact per destination and choose contacts per monitor. Monitors with no list use your default contacts, and a contact can belong to a client workspace.

Each request has an X-Tenium-Signature header with a timestamp and an HMAC-SHA256 signature of the raw body. The docs have a Node example, and you should reject anything older than five minutes.

No. Alerts go to Telegram, chat tools, email and webhooks. If you need a phone call at night, connect PagerDuty and let it page the on-call engineer.

Anything that accepts a webhook already works through the signed webhook contact (Pro) or Zapier (Agency). For a native integration, email us what you use and why; the roadmap page lists what is planned and what is not.

No. You paste an incoming webhook URL that Slack generates for one channel. Tenium can post to that channel and cannot read anything in your workspace.

Yes. Add one Slack contact per webhook URL, then choose contacts per monitor. Monitors without their own list use your default contacts.

Check that the URL starts with hooks.slack.com, that the webhook was not revoked in Slack and that the channel still exists. Then press Send test again.

Yes. Add the bot to the group, then open the connect link inside that group. One contact can serve a whole team, and you can add one group per client.

Reply /ack in the chat or tap the acknowledge link in the alert. Acknowledging tells the team you are on it and stops further escalation steps.

No. The bot only handles the commands sent to it (/start, /ack, /status) and posts alerts to the chat you linked.

No. Email alerts are on every plan, including Free.

So nobody can point alerts at an address that is not theirs. Until the verification link is clicked the contact shows as pending and does not receive alerts.

Yes. Add the client’s address as a separate contact and attach it only to that client’s monitors. Status pages and monthly reports are separate features if you prefer a calmer update.

Yes. Discord, Google Chat and email are available on the Free plan. Slack, Telegram, Teams and Mattermost start at Starter.

Yes, as long as you can create a webhook for it. The URL carries the access, so treat it like a password and delete the webhook in Discord if it leaks.

A site flapping every 30 seconds produces one incident, not one message per blip, because an incident needs confirmation from several regions to open and to close.

The URL of an Incoming Webhook connector for the channel. Tenium requires an https address and posts a message card to it.

Yes. The client name appears in every card, and you can add a separate channel contact per client if you want to keep them apart.

Tenium sends to whatever webhook URL you paste. If your Microsoft 365 tenant no longer offers connectors, use email or the generic webhook (Pro) instead.

No. Google Chat is available on the Free plan together with Discord and email.

Tenium sends a compact text message with the title, the summary and a link. It reads well in a busy space and works in every Workspace edition.

In Google Chat, under the space’s Apps and integrations. If you lose it, create a new webhook, add it as a new contact in Tenium and remove the old one.

Yes, provided the incoming webhook URL is reachable over https from the internet. A server only reachable inside your network cannot receive our posts.

Mattermost starts at the Starter plan, like Slack, Telegram and Teams.

You can, on Pro, if you want the full signed JSON to process yourself. The Mattermost contact is simpler: it posts a readable message with no code on your side.

Read the X-Tenium-Signature header (t=timestamp,v1=signature), compute HMAC-SHA256 of the timestamp, a dot and the raw body with your secret, compare in constant time and reject requests older than five minutes. The docs include a Node example.

A non-2xx answer, a 429, a network error or a 10 second timeout counts as a failed delivery and is retried, up to 3 attempts in total, 30 seconds apart.

Yes, that is the point: the payload has everything needed to create a ticket, post to an internal tool or trigger a script. Zapier (Agency) does the same without code.

Webhooks by Zapier, Catch Hook. Paste its URL into the Zapier contact in Tenium.

Zapier and PagerDuty sit on the top plan, next to client workspaces. On Pro you can reach Zapier yourself with the signed generic webhook if you prefer to build the bridge.

Anything Zapier connects to: create a Jira, Trello or Notion item, send a client email, add a row to a sheet, or post to an app we do not integrate with directly.

Yes. A recovery sends a resolve event with the same dedup key as the trigger, so the PagerDuty incident closes itself.

The integration key (routing key) of an Events API v2 integration on a PagerDuty service. It is not an API token.

No. Tenium has no SMS and no voice calls of its own. If you need phone paging, PagerDuty can do it from the event we send.

Yes. A viewer invited to a workspace sees that workspace and nothing else: not other clients, not billing, not the alert contacts that belong to your team. The login is read-only, so a client cannot create monitors or change alert rules.

Your own contacts are alerted as soon as an incident is confirmed. Contacts that belong to the client are alerted later, after a per-client delay (30 minutes by default). If the monitor recovers before the delay is over, the client alert is skipped, so you get the chance to fix it quietly. Clients without contacts simply see the incident in their view.

Yes. Add a CNAME on the client's DNS and the status page is served on their domain with their logo and colours, without Tenium branding. Custom domains start on Starter and white-label on Pro; on Agency you can have a page for every client.

Uptime for the month, the incident list with durations and causes, response-time trends, certificate and domain expiry dates and missed heartbeats, plus a notes section where you describe the work you did. It carries your logo and the client's, and it is emailed on the 1st to the recipients you set.

No. You invite a client contact as a viewer and they get a login that opens straight into their own workspace, read-only. They never see other clients, billing or your alert contacts, and they never pay us anything.

Yes. White-label is included from Pro ($29) upward, so on Agency you can remove Tenium branding, add your logo and colours, and set a footer such as "Maintained by Your Studio". Agency lets you set branding per client.

It is still generated and sent on the 1st. A report that says zero incidents, certificates valid and response times steady is exactly what a client is paying you to deliver, and you can add a note for work you did.

Write to us through the contact page and we will talk through an arrangement. We would rather hear about it early than have you split clients across two accounts.

No. Plans are monthly, billed in advance, and you can pay yearly and save up to 17%. Cancel from the billing page and the account drops to Free at the end of the period.

Yes. Import from UptimeRobot in one click with a read-only API key, or upload the CSV export. You can assign the imported monitors to one of your client workspaces during the import, and your alert contacts are listed for you to re-add.

Because WordPress renders its own error pages with a success code in several failure modes, the database error screen being the best known. A check that also requires a phrase from the real page turns that silent failure into an alert.

No. The checks run from outside and need nothing installed. A plugin or a one-line mu-plugin is only useful if you want a heartbeat that proves wp-cron itself is executing.

Yes, for an HTTP check on the home page plus a keyword rule that catches the common database-error page. SSL and domain expiry, DNS monitoring and the wp-cron heartbeat start on Starter at $7, so sites that need the full set move up from Free.

It can, because repeated requests to wp-login.php look like probing. Allow the published checker addresses or send a secret header, and read the Cloudflare allow-list guide if your site sits behind it.

With five monitors per site, Starter covers 20 sites, Pro covers 60 and Agency covers 200.

Requests carry a Tenium-Monitor User-Agent, which most analytics tools already filter as a bot. At the fastest interval that is one request every 30 s, plus the occasional confirmation re-check, which is small next to normal traffic.

If the protection is basic auth, enter the credentials in the advanced section on any plan. For cookies or tokens in headers you need Pro, where custom request headers are allowed. Pages that need a multi-step login flow cannot be checked.

Both are cheap. The home page tells you the site is reachable, while a dynamic URL such as a search result or login page exercises the application and database, which a full-page cache can hide.

Up to the timeout it counts as up. To be told about it earlier, use a slow threshold on Pro or lower the timeout; a degraded incident is then raised when the average of the last three checks is over your limit.

Dot notation with optional bracket indexes: status.database, items[0].id or $.data.count. Keys that contain dots or spaces cannot be addressed, so expose a simple alias if you need one.

Not directly, since matching is equality. Make your endpoint return a boolean or a status word that already contains the decision, then assert on that value.

Use a read-only token created just for monitoring, and rotate it like any other credential. The Pro plan accepts any request header, so scope the token to the single route you check.

No. Create one monitor per endpoint. Chained requests and multi-step flows are not part of the product today.

No. Both the page text and your phrase are compared in lower case, so "Add To Cart" and "add to cart" are the same. There is no option to make it case-sensitive.

Not from the page itself, because scripts are not executed. Monitor the JSON endpoint the script reads with the JSON API type, or serve the critical text in the initial HTML.

Create one keyword monitor per site and import them from a CSV with a keyword column; the import guide in the docs lists the columns. Tags make it easy to filter the whole group.

One occurrence is enough for a present check, and any occurrence trips an absent check. There is no counting of matches.

No, they are fixed for every monitor. Each one fires once per certificate, and renewing the certificate starts the sequence again.

Yes, point the monitor at any hostname the wildcard covers. If you use the same wildcard on many sites, one monitor per distinct server is enough because the expiry date is shared.

No. The target must be reachable from the public internet, and a self-signed or private-CA certificate is reported as untrusted by design.

The issuer name is stored and shown on the monitor page, but a change in issuer does not raise an alert by itself. A switch of provider often arrives together with a hostname or chain failure.

Any TLD whose registry publishes RDAP data with an expiration date. Where it does not, the monitor shows a lookup error and no days-left value instead of guessing.

You can enter one, but the registration lookup always runs on the registrable domain above it. Two monitors for sub.clienta.com and www.clienta.com would both read the expiry of clienta.com.

The host has to complete a TLS handshake before the registry step runs, so an HTTPS endpoint must exist on the host you enter. A parked or HTTP-only domain cannot be monitored this way.

Not for this check. Free covers HTTP(S), keyword, ping and port checks, the same basic set UptimeRobot gives away, while SSL and domain expiry start on Starter.

Not as a percentage: each check sends one packet. Intermittent loss appears as scattered failed checks in the history, which the confirmation step keeps from turning into alerts unless regions agree.

Enter an IPv6 address or a hostname that resolves to one and the check uses it. Each check uses a single resolved address, so for dual-stack names monitor the explicit address you care about.

The checker environment was not permitted to send ICMP, so it tried TCP ports 443 and 80. The check details say which method was used.

No. Private and loopback ranges are rejected before any packet is sent, since the checkers cannot route to your network.

No, the monitor only completes the TCP handshake and closes the connection. It will notice a stopped mail server but not one that accepts connections and then misbehaves.

Ping asks the operating system whether the host answers, while a port check asks whether a specific service is listening. A host can pass one and fail the other, which is why pairing them quickly narrows down a failure.

The checkers will time out. Allow the published checker addresses, or monitor something that is already public instead of opening a private service.

They count toward your plan's total monitors: 30 on Free, 100 on Starter, 300 on Pro and 1000 on Agency.

The job has to make one outbound HTTPS request, so a network-isolated host cannot report in directly. Let a neighbouring machine that can reach us call the heartbeat after reading a status file from the isolated one.

Yes, as a monitor of type heartbeat, with a period and grace. The API documentation shows the request, and write access requires the Pro plan or above.

The random token in the URL is the credential. If it leaks, delete the monitor and create a new one so that the old URL stops working.

No. A heartbeat is one monitor against your plan total, and the number of pings you send does not change anything, apart from the rate limit.

Yes, enter the full name, for example _dmarc.clienta.com or shop.clienta.com, and pick the record type that name has.

It tells you when the answer stops containing your expected values, or when the record disappears. It does not store a before-and-after history of arbitrary changes, so list the values you care about.

Validation is done by the public resolvers. A zone with broken DNSSEC typically comes back as SERVFAIL, and that is reported as a failure.

Not at present; queries go to the two public resolvers. A UDP monitor on port 53 tells you whether your authoritative server answers at all.

No. Every monitor works from the outside through public URLs. The only thing you add is a line in the server's scheduler for the cron heartbeat, if you want one.

Pages that are public and cheap to render: the home page, /wp-login.php and /?rest_route=/. Avoid /wp-admin, which redirects to the login for anonymous users, and never include credentials in a monitored URL.

Not as a one-click product. The set above is quick to create by hand or by CSV import, and tags keep the group together.

Treat each site in the network as its own hostname and monitor them separately, plus one REST check on the main site. A network-wide failure then shows up as many incidents at once.

No. Checks are single requests that cannot carry a cart between steps, so there is no simulated purchase. The monitors prove that each page and the layers beneath it work, which catches the large majority of breakages.

Yes for page-level checks, but the block checkout renders much of its content with JavaScript, so choose a phrase that appears in the server-delivered HTML, or lean on the Store API and cart monitors.

Caching affects the product and home pages, which is why the Store API and cart checks matter. They run PHP and touch the database even when public pages are served from a cache.

Keyword checks are on every plan, so the free plan can already watch product, cart and checkout pages. JSON checks need Starter, which also unlocks SSL and domain expiry. Pro adds the slow-checkout threshold, escalation and custom request headers.

Not today. The probe is chosen by port and any reply is accepted. If you depend on a protocol-specific handshake, monitor through a TCP port, a status page or a heartbeat instead.

A REFUSED or SERVFAIL response still proves the server is alive and answering UDP 53. To check that names resolve correctly, add a DNS monitor with expected values.

Java Edition servers listen on TCP, so a port monitor works. Bedrock and most Steam-based games use UDP with their own query formats, which the generic probe does not speak.

Three to five seconds suits resolvers and NTP. Longer values only delay the verdict, and anything above 15 seconds is reduced to 15 anyway.

You get the first 600 characters of the new text and the hash fingerprints, not a line-by-line comparison. Keep your own copy of the page if you need to see exactly which words moved.

The new content becomes the baseline immediately, so you receive a single alert. The incident closes on the next check that sees the same content.

Binary files are read as text, which makes the hash fragile and the excerpt unreadable. It will notice a replaced file, but this is not a feature built for documents.

There are no ignore rules. Use a selector for the stable part of the page, or point the monitor at a cleaner source of the same data.

No. It is the same check under the name many people search for. The port monitoring page covers the same feature with database examples.

Not at the TCP level. It proves the port accepts connections. For protocol-level proof use an HTTP, keyword or DNS check, or a heartbeat sent by the service itself.

No, UDP has no handshake. Use the separate UDP monitor, which sends a probe and waits for any reply.

No. Private ranges are refused because the checker regions cannot reach them. Use a heartbeat that the internal service sends to Tenium instead.

A slow response opens a degraded incident on Pro and Agency when the slow threshold is exceeded. Whether that counts toward your SLA is a contract question; the incident list gives degraded incidents their own severity, so you can tell slowness from outages.

The calculator on this page does the arithmetic, and the report shows the measured figure. There is no automatic breach alert today, so compare the report with your target on the 1st.

Yes. Alerts are suppressed during a window and the period is excluded from reported uptime. Schedule windows before the work starts.

Free keeps 90 days, Starter and Pro 365 days, and Agency 730 days.

It is a beta feature and the app does not charge for it separately. The plan table above shows what each plan includes for the monitors you link.

Not today. The catalogue is fixed. If you rely on a service that is missing, tell us through the contact page.

No. It reads the feed. Your own monitors will usually notice first, and the tag appears on those incidents afterwards when the feed catches up within the window.

Only monitors linked to that provider are tagged, and only for incidents that open within 30 minutes after the feed turns non-operational.

No. The stored figure is the time until the body has arrived. A time to first byte would be shorter and would hide a slow body.

No. Times are stored and charted on every plan for 5 min checks and faster. Only the threshold alert and the degraded incident need Pro or Agency.

One slow request is noise. Three in a row averaging above your limit is a trend. It trades a minute or two of delay for far fewer false alarms.

It is recorded as a separate degraded incident, distinct from an outage, so you can decide how your contract treats it.

No. We have not shipped one. This page exists so the plan is stated plainly instead of implied. Nothing on it is available to install, and we will not announce a date until we have one.

The REST API. An MCP server would be a thin layer that exposes a few of the API's endpoints as tools an assistant can call, using an API key you create and can revoke. Anything it could do, you could already do with curl.

Reading status and incidents uses a read key, which every plan has. Anything that changes something, such as creating a maintenance window or acknowledging an incident, needs write access, which is on Pro ($29) and Agency ($59).

Use the contact page and mention MCP and the assistant you use. A short note about the job you would hand over helps us pick the first tools.

It can check the endpoint the form posts to, and that the page still contains the form text. It cannot submit a real lead, so keep one manual test in the launch checklist.

No. Analytics measures behaviour. Monitoring measures whether the page and the endpoints behind it were reachable and correct.

Yes. Add a Slack, Telegram or e-mail contact that the team shares. Escalation to a second contact is on Pro.

Yes, through the public status page and through alerts sent to your support channel. Extra seats on the account depend on the plan.

No. Components turn red and green by themselves, and notes and incident text are written by you.

No. It alerts your team and updates a page. There are no SMS or voice notifications to customers.

Follow the provider in the third-party feature (beta). While it is non-operational, incidents on linked monitors carry its name.

You choose the regions per monitor from the ones that are live and allowed by your plan. We recommend leaving confirmation on for any public site, since it needs several regions to work.

The other regions answered successfully, so the failure was treated as local to one path and no incident opened. The failed result is still stored in the monitor history, so you can see it afterwards.

Paid plans use every live region up to their allowance, so new locations apply automatically. Free stays at 1 region.

They are published at /checker-ips and as JSON, and the list is what to use for firewall and WAF allow-lists. Re-fetch it from the JSON endpoint instead of copying addresses into a rule once and forgetting.

No. We measure how long the server and network take to deliver the response, not how long a browser takes to render it. Use it to catch hosting and application slowness, and a lab tool for front-end work.

Pro and Agency. The timing charts and history are visible on every plan, and the alert rule is what needs the upgrade.

Set the threshold from your own baseline, use the p95 rule rather than single samples, and let quiet hours hold non-critical alerts overnight. A recovery message arrives when the p95 drops back under the line.

You can monitor any URL you have the right to check, so yes in practice, but we do not publish benchmarks and the numbers only mean something from the same regions and at the same interval.

No. Custom domains start on Starter ($7 per month). Free pages are served from a Tenium address with Tenium branding.

Yes. Incidents open and close from monitor state, and the daily bars and uptime figures are recalculated automatically. You only write by hand when you want to add context to an incident or announce maintenance.

Email subscribers receive an update when an incident resolves, and RSS readers see it in the feed. Contacts you configure for alerts get a separate recovery notification that includes the downtime duration.

Features tied to a higher plan, such as white-label branding, are no longer available, and you should bring the number of pages down to the new limit first. The limits for every plan are listed on the pricing page.

Yes. Generate it on demand for the month you want and send it to yourself first. Add your notes before the 1st so the automatic version already contains them.

No. The PDF arrives by email. If you also invite them as viewers on an Agency plan, they can open the same reports inside their workspace.

On Agency you set the client’s logo on the cover and keep your own footer. White-label removes Tenium mentions on Pro and above.

The report is still generated. It shows 100% or the measured uptime, an empty incident table and the certificate, domain and response-time sections, which is often the most reassuring report of all.

The practical limit is the monitor allowance, 1000 on Agency. The plan defines no separate cap on the number of workspaces.

No. A viewer sees only their own workspace, and the workspace list, billing and your team’s alert contacts are hidden from them.

A viewer login is scoped to one workspace. If a person looks after two entities, invite them to each, or make them a team member if they should see everything.

No. Pro includes white-label status pages and one branded report, but client workspaces and per-client reports are Agency only. If you outgrow Pro, the upgrade keeps your monitors and history.

No. Free and Starter pages carry Tenium branding; Starter adds a custom domain. Removing branding and adding your own logo and footer is a Pro feature.

Not when white-label is on. The cover, footer and header use your assets. Report data comes from the same incidents shown in the app.

Yes, on Agency, where each client workspace has its own branding. On Pro the branding is account-wide.

Not if you use a custom domain. Add a CNAME for a subdomain of the client’s own domain and the page, and its certificate, are served from there.

No. Acknowledging records that someone is handling it and stops further escalation steps. The incident closes when the monitors confirm that the site has recovered.

Yes. Add notes during or after the incident, in the app or through the API. They stay in the timeline and are included in the monthly report for that month.

The site responded, but with a challenge or block page aimed at bots. It is shown separately from down so you can tell a WAF rule from a real outage; the fix is usually to allow-list the checker IPs.

Yes, with confirmation, evidence and acknowledgement. Escalation steps are the part that needs Pro or above.

If the window is set to announce, it appears as upcoming maintenance on the page, and email subscribers can be notified. Switch announcements off for internal work that clients should not see.

By default yes: checks stop for the duration. If you turn on keep checking, checks continue, alerts and incidents stay suppressed, and the downtime that happens is recorded against the window.

Yes, by leaving the exclude-from-uptime switch on, which is the default. Turn it off for a window if the contract counts that time against you.

When the window ends, normal monitoring resumes. If the site is still down at that point, a confirmed failure opens an incident and alerts go out as usual. Extend the window before it ends if you know it will overrun.

No. Telegram alerts start on Starter. Free includes email, Discord and Google Chat.

No. You connect to the Tenium bot with a one-time link. There is no BotFather step and no token to paste.

Yes. Reply /ack in the chat, or tap the acknowledge link in the alert. Either one stops further escalation steps for that incident.

Delete the contact in Integrations and alerts stop going there. Removing the bot from the group on the Telegram side has the same practical effect.

Pro and Agency. Starter and Free alert every contact straight away, with no delay between steps.

No. It cancels the later escalation steps. The people who already heard about the incident still receive a recovery message when the site is back.

From the moment the incident opens, which is when the failure has been confirmed by the other regions. Confirmation takes up to about a minute by default, so the first step is not instantaneous.

Yes. The delay belongs to the contact and each monitor chooses its own contacts, so a different chain per client means different contacts on that client’s monitors. On Agency, client workspaces also have their own default contacts.

Yes, and some people do exactly that. Nothing in either product requires exclusivity, and a split setup also gives you a second opinion on outages. Just remember that alerts then arrive from two places.

No. The import keeps each interval as it was and only enforces the floor of your Tenium plan. Change them in bulk afterwards if you move to a plan that allows faster checks.

No. There is no separate discount or bonus for importing. The import itself is free and shows a preview before anything is created.

On published prices, yes for a team of three at 100 monitors, by the percentage shown in the price section above. For one or two people the entry tiers compare differently: Tenium Starter at $7 per month is about 71% cheaper than UptimeRobot Solo (50 monitors) at €22 (about $24) per month.

No, it is our arithmetic on their public per-unit prices for 100 monitors and three responders. Your invoice depends on the exact plan, billing period and any negotiated terms.

Not if you run rotations. Tenium can escalate from one contact to the next by Telegram and email, but it does not manage schedules, swaps or overrides.

They stay in Better Stack. Tenium does not execute browser scripts, so you either keep that part there or reduce the check to an HTTP or keyword assertion.

No. Most teams move client websites first and keep engineering systems where they are. Both tools can alert the same chat channel while you compare.

Because the tier that matches 100 checks is not on their public price list as far as we can verify. We would rather show nothing than a number you cannot reproduce.

No. Tenium does not run scripted browser flows, so these cannot be converted. Replace the important ones with keyword or JSON checks against the endpoints they exercise.

Historic reports stay in Pingdom. Export the PDFs or screenshots your clients may ask for before you cancel, because importing the list brings monitors across, not past results.

Pro includes one branded monthly PDF with uptime, incidents and response times, and Agency sends one per client automatically. It is a monthly report rather than an on-demand SLA calculator.

For 100 tests at one-minute checks and up to 2 people, no: Tenium Starter ($7) costs less than Superior on list prices. For a team of three, Superior costs less than Tenium Pro, and Pro is the plan once you need white-label or 30-second checks.

Uptime, keyword, ping and port tests do. Page-speed tests and the server agent have no counterpart, so leave them out of the list or untick them in the preview.

Only if it is on your own domain. Point the CNAME at the target Tenium shows; the old vendor-hosted URL stops being yours when you cancel.

Starter has 2, Pro 5, and Agency is unlimited. StatusCake limits depend on plan, so check their pricing page for the tier you use.

According to their public plans, yes: white-label sits on the $249 tier. Check their pricing page, because plan contents change.

Less so. You can set logo, favicon, colours, footer text and a custom domain, plus component groups, but the layout options are narrower than Hyperping.

Yes. Point only the monitors, not the public domain, at Tenium first, compare incidents for a week, and then move the CNAME.

Single sign-on is not part of any plan described on this site. If it is a hard requirement, that is a reason to stay with a provider that has it.

Plan on about five: home page, keyword, certificate, domain and one more. There is no one-click template, so create them by hand or as CSV rows.

No. If link health is part of your service to clients, keep Oh Dear for that and use Tenium for uptime, certificates and reporting.

Then per-site pricing is probably fine for you, and any difference is unlikely to justify switching tools. Compare their tier for your exact site count before deciding.

On Agency, yes. A client workspace has viewer logins that open directly to that client and cannot see any other workspace.

No. A heartbeat is a plain HTTP request, so any language that can make a GET or POST can report in. It is simpler and less rich.

It is the extra time after the expected run before the heartbeat is treated as missed. Set it longer than your slowest normal run to avoid false alarms.

In Tenium they have their own allowance (100 on Pro), separate from the 300 monitors. In Cronitor each heartbeat is a monitor you pay for.

Yes, up to 10 KB of text in the body or query of the ping, which appears in the incident. It is a short log tail, not full telemetry.

Not necessarily. Checkly's starting price is lower than Tenium Pro, but it adds run-based usage, which depends on how many checks you run and how often.

You can create them through the REST API, with write access on Pro, from any script. There is no declarative config format or Terraform provider.

It can check the endpoints behind them with JSON assertions or request headers, but it does not drive a browser. A broken button that returns a normal response will not be noticed.

Yes, nothing is changed at the source. Run both until you have seen a real incident handled by Tenium.

In cash, yes: a small server is cheaper than any paid Tenium plan. The comparison on this page also counts your time, backups and the lack of a second location.

No, checks come from the public internet. For anything behind a firewall, run something inside the network, such as Uptime Kuma, and send heartbeats out if you want Tenium to notice when it stops.

Yes, and it is a sensible setup. A heartbeat from your internal monitor to Tenium tells you when the monitor itself has died.

No. Tenium enforces the minimum for your plan: 1 min on Starter, 30 s on Pro and Agency.

Their tiers depend on subscriber numbers and features, and we do not have one verified figure for your case. Look at their pricing page and add a monitoring tool.

Email and RSS. Your own team alerts go to Telegram, email and the chat tools listed on the integrations page.

Yes. Use Tenium to detect the outage and notify your team by webhook, then post to Statuspage if you want to keep it as your public channel.

Yes, daily bars with a 30 and 90-day uptime figure, with maintenance windows excluded from the calculation.

No. It never logs in to your sites. It checks them from outside and reports on availability, response time, certificates and domain expiry.

Not for availability alerts. Some agencies keep it briefly as a cross-check and drop it once they trust the new setup.

If you keep both active, yes, for the same outage. Send them to different channels during the overlap, or switch one off for the sites you have moved.

There is no one-click template. The usual five checks (home page, keyword, certificate, domain, one more) are created by hand or from CSV rows, and the import preview lets you edit them before anything is created.

Their pricing page shows "starting at $9" a month and a quantity selector; the text we read on 2026-11-08 does not say how many checks the starting price includes. Set the selector to 100 on their page for the real figure. Tenium Starter is $7 a month for 100 monitors.

The pages we read mention no free plan or trial, only a 30-day money-back guarantee for new users. Tenium Free gives 30 monitors with 5 min checks.

No on both. Tenium checks single URLs, keywords, JSON paths, ports, DNS, certificates and heartbeats, and alerts by email, Telegram, Slack, Discord, Teams and webhooks. If you need scripted flows or phone alerts, stay with a tool that has them.

Upload a CSV with a name and a URL column, or paste a list of URLs; you see a preview before anything is created. We have not tested Uptime.com's own export format, and alert contacts and status pages are not imported.

For about 20 monitors and one user, Site24x7 Web Uptime at $10 a month is close to Tenium Starter at $7. For 100 monitors the add-on packs bring it to about $46 by our arithmetic against $7 for Tenium Starter, which covers 100 monitors with 1 min checks.

The pages we read offer a 30-day full-feature trial. Tenium Free has no time limit and gives 30 monitors with 5 min checks.

No. Tenium checks websites, APIs, ports, DNS, certificates, domains and heartbeats from outside. It has no server agent and no application monitoring.

Agency ($59, 1000 monitors). Each client gets a workspace with a read-only viewer login and a branded monthly report sent on the 1st.

Look at the rows. If every region reports a success, the site is reachable from the outside and the cause is on your side: your network, resolver, VPN, a hosts file entry or a browser extension. If every region fails with the same error, the site is down for everyone. Mixed rows point at a CDN or routing problem that affects some networks only.

Because the request comes from a data centre and carries no browser fingerprint. Cloudflare, Akamai and similar services often serve a challenge or a block page to such traffic. The site is up; the result only tells you that bots are filtered. A monitor can send a secret header that your firewall recognises, which is covered on the HTTP headers page.

It runs from the regions that are live today: Amsterdam and London. If the check service answers from only one of them, you are looking at one vantage point and not at a confirmed outage. Tenium's own monitoring is the place where a failure is verified from the other regions before anyone is alerted.

No. The request is anonymous, public and made without cookies, and addresses in private ranges such as 10.x.x.x or 192.168.x.x are rejected. For login-protected pages use a monitor with basic auth or a custom header instead.

Yes, a modest per-visitor and global rate limit applies so the tool stays free and cannot be used to hammer someone's server. If you hit it, wait a minute. Continuous checking is what a monitor is for.

43.2 minutes in a 30 day month, 10.1 minutes per week, 86 seconds per day and 8.76 hours per year. In a 31 day month the allowance is 44.6 minutes.

Rarely. It allows 4.3 minutes a month, less than most hosts need to restart a database. It needs redundancy at every layer plus fast automated failover. Most small sites are better served by a promise of 99.9% together with quick detection.

That depends entirely on the contract. Many SLAs exclude announced maintenance, some cap it at a number of hours per month. In Tenium, maintenance windows are excluded from the uptime figure, so it matches an SLA that carves them out.

A host usually measures its own network or servers, while an external monitor measures the full path a visitor takes, including DNS, TLS and the application. Application errors, expired certificates and slow responses that exceed a timeout count only in the second figure.

Multiply the availability of components that are all required, as in the stacking section above. For redundant components, multiply the failure probabilities and subtract from one, assuming the failures are independent.

Every 15 minutes (at :00, :15, :30, :45) during hours 9 through 17, Monday to Friday. The final run each day is at 17:45. That is 36 runs per weekday and none on Saturday or Sunday.

Standard cron combines the two day fields with OR. If both are restricted, the job runs when either matches. To get "the first Monday", run the job on days 1-7 and test the weekday inside the script.

It uses the time zone of the machine running the daemon, often UTC on hosted servers. The preview shows the next runs in your browser's zone, so compare it with what the server's date command prints.

No. It handles the classic five-field Unix format only. Quartz adds a seconds field and special characters, and AWS EventBridge uses six fields with a year and a required question mark.

Add a request to a heartbeat URL after the command, and give the heartbeat a period and a grace time. If no ping arrives before the deadline, an incident opens. The curl line generated under the preview shows the pattern.

Enter the hostname above and read the Days left value, or run echo | openssl s_client -servername example.com -connect example.com:443 2>/dev/null | openssl x509 -noout -dates in a terminal, which prints the notBefore and notAfter dates.

The verdict summarises dates. The browser also checks that the chain is complete and that the name matches. A hostname mismatch or an untrusted issuer appears in the HTTPS status cell, and an openssl run shows the full chain.

With 90 day certificates, renew at about 60 days, leaving 30 days to fix failures. If you renew manually, make 30 days remaining the trigger. Any certificate with under 14 days left and auto-renewal supposedly on has a broken renewal.

Not directly. It reports the leaf certificate's expiry and issuer. A missing intermediate usually surfaces as an error in the HTTPS status cell, but for the exact chain use openssl s_client with -showcerts.

No. It connects to the standard HTTPS port on public hostnames only. An SSL monitor in Tenium lets you choose another port for services that use TLS from the first byte.

At minimum Strict-Transport-Security, X-Content-Type-Options: nosniff, a Referrer-Policy and a Content-Security-Policy (in report-only mode first). Cache-Control matters for performance, and cookies need Secure and HttpOnly.

The check service may not return them for every request, or the request failed before a response arrived. In that case use curl -sI against the URL from your own machine.

Look for cf-ray, x-cache, x-served-by or via. A numeric Age header above zero means a cached copy was delivered. If every request shows MISS, check Cache-Control and Vary.

Yes, up to five. The headers shown are those of the final response. To inspect each hop, run curl -sIL and read the Location lines.

The request is a plain GET without cookies or browser hints from a data centre address. Bot rules, geographic rules and device detection can all change the response. That is the same view a monitor has, so it is also the one worth testing.

Enter it above to read the expiration event from RDAP, or log in at the registrar. If the result says "not published", the registry hides expiry and the registrar account is the only source.

It covers the same ground in a modern, machine-readable form. RDAP gives JSON and standard status codes, and many registries are retiring plain WHOIS for generic domains. Personal contact data is generally redacted in both.

Renew it at the registrar immediately, before the redemption stage if you can. In redemption you must pay a restore fee, and once the state is pendingDelete you can only wait. The status flags in the result show where it is.

Some registries return no registrar entity in public RDAP, or return a reseller. Reseller or privacy-service names can appear instead of the company you pay.

Not here: the page handles one lookup at a time. A set of SSL / domain monitors does the same continuously for a list, and the monitor page works as a renewal calendar.

A /24 has 254 usable hosts out of 256 addresses, a /26 has 62 out of 64 and a /28 has 14 out of 16. Subtract the network and broadcast addresses from the total.

They are bitwise opposites. 255.255.255.192 as a subnet mask has the wildcard 0.0.0.63. Subnet masks mark the network bits; wildcards mark the bits that may vary and are used in access lists and routing protocol statements.

A /32 is a single address and a /31 is a two-address link where both ends are usable under RFC 3021, so neither reserves addresses for network and broadcast.

Enter each with the same prefix length and compare the network addresses. If they match, the hosts are in the same block. 10.0.5.200/23 and 10.0.4.9/23 are both in 10.0.4.0/23, but 10.0.6.9/23 is not.

Under 200 ms for cached or static pages and under 500 ms for dynamic ones, measured from where your visitors are. Web performance guidance treats a time to first byte up to 800 ms as good. Over a second needs investigation.

Distance sets a minimum round-trip time, and every new connection needs several round trips. A CDN, HTTP/3 and TLS session reuse reduce the effect, but a site served from a single data centre will always look slower to far-away regions.

No. It stops when headers arrive and downloads no body, images, scripts or stylesheets. For full page load use a browser tool such as Lighthouse; this test tells you about the server and the network.

Cache state, DNS caching, connection reuse and server load all vary. Repeat the test and look at the range. Continuous measurement reduces the noise by averaging and by showing percentiles.

The page has a row for each live region (2 regions today), but a single call may be answered by one of them. Monitoring in Tenium measures from every enabled region on each check.

No. It is a static file with the text you entered. For a figure that follows the data, generate it with a script from monitoring data or link to a live status page.

GitHub typically removes data: image URLs in Markdown. Download the SVG, commit it to the repository, and reference it by relative path or by its raw URL.

Yes, the SVG is yours to use. Show a figure you have measured and name the period, because a client may later check it against the report.

It is 20 pixels tall with a width based on the text, and as an SVG it scales without loss. Set width or height on the img tag to resize it.

Thirty days is current and shows quick improvement, while 12 months hides short bad stretches. Whichever you choose, write it in the label and keep it consistent.

Resolvers cache independently, and each keeps its copy until the TTL runs out. Right after a change one resolver may show the new value and another the old. Anycast and geo-aware DNS can also deliberately return different addresses by location.

An A record maps a name to IPv4 addresses. A CNAME says "this name is an alias of another name" and the resolver follows it to that name's records. A name with a CNAME cannot hold other record types.

As long as the previous TTL, since that is how long caches may keep the old value. With a TTL of 3600 expect up to an hour; with 86400 up to a day. Lower the TTL in advance for planned moves.

Only records that exist are shown. A domain without IPv6 has no AAAA, a CNAME hides the others, and many domains have no CAA, which means any authority may issue certificates.

They are TXT records at specific names and have their own pages under /tools: the SPF, DMARC and DKIM lookups. The TXT result here covers the apex name only.

It is the preference value. Senders try the lowest number first and move to higher numbers only if that host is unreachable. The absolute values do not matter, only their order; 10 and 20 behave like 1 and 2.

Either no mail service is configured, the records are on a different name (check the exact domain or subdomain you use in addresses), or you are seeing a cached empty answer. Without any MX, senders fall back to the A record.

Not to an IP address: it must be a host name. Pointing to a CNAME is forbidden by the standard and works only with tolerant mail servers, so use the real host name.

The MX lookup only proves where mail would go. Next, resolve each listed host to an address, confirm that port 25 accepts a connection from outside and send a message from another provider. SPF, DKIM and DMARC then decide whether it lands in the inbox.

Usually not. Senders already queue and retry for days. A backup that accepts mail without checking recipients adds spam and backscatter. If you want one, make sure it filters like the primary.

No. SPF, DKIM and DMARC establish who you are; they do not decide whether a message is wanted. Content, sending reputation and engagement still decide the folder. Failing them, on the other hand, reliably costs deliverability.

Nowhere except to Cloudflare's public DNS resolver, directly from your browser, which is the same as typing the domain into any DNS lookup. Tenium does not receive the domain you type.

-all is stricter. With DMARC at quarantine or reject, ~all is a common and reasonable choice, because DMARC does the enforcing and ~all avoids rejecting forwarded mail outright. Either is acceptable; +all and ?all are not.

Remove services you no longer use, replace include records with their literal ip4 and ip6 ranges where the provider's addresses are stable (called flattening), or move some senders to a subdomain with its own SPF record.